diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.16124291412975017,
+  "eval_steps": 500,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.788908765652952e-08,
+      "loss": 7.7715,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.577817531305904e-08,
+      "loss": 7.8519,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3667262969588554e-08,
+      "loss": 8.4059,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.155635062611808e-08,
+      "loss": 7.562,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.94454382826476e-08,
+      "loss": 7.5777,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0733452593917711e-07,
+      "loss": 7.6764,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2522361359570663e-07,
+      "loss": 7.9352,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4311270125223616e-07,
+      "loss": 8.0042,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.610017889087657e-07,
+      "loss": 7.8412,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.788908765652952e-07,
+      "loss": 7.8339,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9677996422182471e-07,
+      "loss": 7.8978,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1466905187835422e-07,
+      "loss": 7.3701,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3255813953488374e-07,
+      "loss": 8.0251,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5044722719141327e-07,
+      "loss": 7.7323,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6833631484794277e-07,
+      "loss": 7.6358,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.862254025044723e-07,
+      "loss": 7.9594,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.041144901610018e-07,
+      "loss": 7.6372,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.220035778175314e-07,
+      "loss": 8.1336,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.398926654740609e-07,
+      "loss": 7.5629,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.577817531305904e-07,
+      "loss": 7.6304,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7567084078711993e-07,
+      "loss": 7.7794,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9355992844364943e-07,
+      "loss": 8.2466,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1144901610017893e-07,
+      "loss": 7.4072,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2933810375670843e-07,
+      "loss": 7.7937,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.47227191413238e-07,
+      "loss": 7.8197,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.651162790697675e-07,
+      "loss": 7.4801,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.830053667262969e-07,
+      "loss": 7.3983,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.008944543828265e-07,
+      "loss": 7.4355,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.18783542039356e-07,
+      "loss": 7.8755,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.366726296958855e-07,
+      "loss": 7.8491,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.545617173524151e-07,
+      "loss": 7.4471,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.724508050089446e-07,
+      "loss": 7.6026,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.90339892665474e-07,
+      "loss": 7.5104,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.082289803220036e-07,
+      "loss": 7.179,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.26118067978533e-07,
+      "loss": 7.3247,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.440071556350628e-07,
+      "loss": 7.1524,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.618962432915923e-07,
+      "loss": 7.3259,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.797853309481218e-07,
+      "loss": 7.1828,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.976744186046513e-07,
+      "loss": 7.1282,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.155635062611808e-07,
+      "loss": 7.1449,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.334525939177103e-07,
+      "loss": 7.2439,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.513416815742399e-07,
+      "loss": 7.0199,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 7.497,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.871198568872989e-07,
+      "loss": 7.2569,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.050089445438284e-07,
+      "loss": 7.0604,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.228980322003579e-07,
+      "loss": 6.915,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.407871198568874e-07,
+      "loss": 6.7001,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.586762075134169e-07,
+      "loss": 6.6596,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.765652951699465e-07,
+      "loss": 6.4777,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.94454382826476e-07,
+      "loss": 6.5302,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.123434704830055e-07,
+      "loss": 6.3794,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.30232558139535e-07,
+      "loss": 6.5417,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.481216457960645e-07,
+      "loss": 6.6661,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.660107334525939e-07,
+      "loss": 6.676,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.838998211091236e-07,
+      "loss": 6.422,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.001788908765653e-06,
+      "loss": 6.5479,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0196779964221826e-06,
+      "loss": 6.5683,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.037567084078712e-06,
+      "loss": 6.48,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0554561717352416e-06,
+      "loss": 6.4804,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.073345259391771e-06,
+      "loss": 6.4867,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0912343470483006e-06,
+      "loss": 6.2915,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1091234347048303e-06,
+      "loss": 6.3045,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1270125223613598e-06,
+      "loss": 6.1512,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1449016100178893e-06,
+      "loss": 6.1851,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1627906976744188e-06,
+      "loss": 6.2033,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.180679785330948e-06,
+      "loss": 6.1867,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1985688729874776e-06,
+      "loss": 6.2157,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2164579606440073e-06,
+      "loss": 5.9342,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2343470483005368e-06,
+      "loss": 5.8863,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.252236135957066e-06,
+      "loss": 5.6881,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2701252236135958e-06,
+      "loss": 5.8519,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2880143112701255e-06,
+      "loss": 5.9884,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3059033989266548e-06,
+      "loss": 5.7854,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3237924865831845e-06,
+      "loss": 5.8981,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3416815742397138e-06,
+      "loss": 5.7547,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3595706618962435e-06,
+      "loss": 5.8039,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3774597495527728e-06,
+      "loss": 5.737,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3953488372093025e-06,
+      "loss": 5.7724,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.413237924865832e-06,
+      "loss": 5.5584,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4311270125223615e-06,
+      "loss": 5.5668,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.449016100178891e-06,
+      "loss": 5.7735,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4669051878354205e-06,
+      "loss": 5.5887,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.48479427549195e-06,
+      "loss": 5.7798,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5026833631484797e-06,
+      "loss": 5.4861,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.520572450805009e-06,
+      "loss": 5.4885,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 5.7567,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.556350626118068e-06,
+      "loss": 5.4566,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5742397137745977e-06,
+      "loss": 5.3949,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.592128801431127e-06,
+      "loss": 5.518,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6100178890876567e-06,
+      "loss": 5.5435,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6279069767441862e-06,
+      "loss": 5.4049,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6457960644007157e-06,
+      "loss": 5.5404,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6636851520572452e-06,
+      "loss": 5.5624,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6815742397137747e-06,
+      "loss": 5.4432,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6994633273703042e-06,
+      "loss": 5.398,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7173524150268337e-06,
+      "loss": 5.2518,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7352415026833632e-06,
+      "loss": 5.3261,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.753130590339893e-06,
+      "loss": 5.057,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7710196779964222e-06,
+      "loss": 5.4077,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.788908765652952e-06,
+      "loss": 5.4552,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8067978533094812e-06,
+      "loss": 5.5213,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.824686940966011e-06,
+      "loss": 5.3905,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8425760286225404e-06,
+      "loss": 5.2055,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86046511627907e-06,
+      "loss": 5.1314,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8783542039355994e-06,
+      "loss": 5.2104,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.896243291592129e-06,
+      "loss": 5.2958,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9141323792486586e-06,
+      "loss": 4.9671,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9320214669051877e-06,
+      "loss": 5.1641,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9499105545617176e-06,
+      "loss": 5.0792,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.967799642218247e-06,
+      "loss": 5.2797,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9856887298747766e-06,
+      "loss": 4.9679,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.003577817531306e-06,
+      "loss": 4.9545,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0214669051878356e-06,
+      "loss": 4.9588,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.039355992844365e-06,
+      "loss": 4.9073,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0572450805008946e-06,
+      "loss": 4.8583,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.075134168157424e-06,
+      "loss": 4.8729,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0930232558139536e-06,
+      "loss": 4.9038,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.110912343470483e-06,
+      "loss": 4.7961,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1288014311270127e-06,
+      "loss": 4.9145,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.146690518783542e-06,
+      "loss": 4.7004,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1645796064400717e-06,
+      "loss": 4.7949,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.182468694096601e-06,
+      "loss": 4.8802,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2003577817531307e-06,
+      "loss": 4.8386,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2182468694096606e-06,
+      "loss": 4.834,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2361359570661897e-06,
+      "loss": 4.5009,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2540250447227196e-06,
+      "loss": 4.7804,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2719141323792487e-06,
+      "loss": 4.5393,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2898032200357786e-06,
+      "loss": 4.3516,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 4.8557,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3255813953488376e-06,
+      "loss": 4.5584,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.343470483005367e-06,
+      "loss": 4.5828,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.361359570661896e-06,
+      "loss": 4.7232,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.379248658318426e-06,
+      "loss": 4.7613,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.397137745974955e-06,
+      "loss": 4.4471,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.415026833631485e-06,
+      "loss": 4.6829,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4329159212880146e-06,
+      "loss": 4.601,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.450805008944544e-06,
+      "loss": 4.3626,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4686940966010736e-06,
+      "loss": 4.6798,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.486583184257603e-06,
+      "loss": 4.4511,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.504472271914132e-06,
+      "loss": 4.389,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.522361359570662e-06,
+      "loss": 4.2783,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5402504472271916e-06,
+      "loss": 4.2079,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.558139534883721e-06,
+      "loss": 4.2643,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.576028622540251e-06,
+      "loss": 4.6845,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.59391771019678e-06,
+      "loss": 4.5027,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6118067978533096e-06,
+      "loss": 4.2404,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.629695885509839e-06,
+      "loss": 4.3328,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.647584973166369e-06,
+      "loss": 4.0663,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6654740608228985e-06,
+      "loss": 4.5931,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6833631484794276e-06,
+      "loss": 4.214,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.701252236135957e-06,
+      "loss": 4.5132,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.719141323792487e-06,
+      "loss": 4.158,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7370304114490165e-06,
+      "loss": 4.4817,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7549194991055456e-06,
+      "loss": 4.2528,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.772808586762075e-06,
+      "loss": 4.208,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.790697674418605e-06,
+      "loss": 4.3172,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8085867620751345e-06,
+      "loss": 4.4963,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.826475849731664e-06,
+      "loss": 4.142,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.844364937388193e-06,
+      "loss": 4.3967,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.862254025044723e-06,
+      "loss": 4.4727,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8801431127012525e-06,
+      "loss": 4.4704,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.898032200357782e-06,
+      "loss": 4.2377,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.915921288014312e-06,
+      "loss": 4.2584,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.933810375670841e-06,
+      "loss": 4.1697,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9516994633273705e-06,
+      "loss": 4.3103,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9695885509839e-06,
+      "loss": 4.1562,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.98747763864043e-06,
+      "loss": 4.1238,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0053667262969594e-06,
+      "loss": 4.3442,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0232558139534885e-06,
+      "loss": 4.4275,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.041144901610018e-06,
+      "loss": 4.3718,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0590339892665475e-06,
+      "loss": 4.1011,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 4.1713,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0948121645796065e-06,
+      "loss": 4.1864,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.112701252236136e-06,
+      "loss": 4.1596,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1305903398926655e-06,
+      "loss": 4.3598,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1484794275491954e-06,
+      "loss": 4.3394,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.166368515205725e-06,
+      "loss": 4.2193,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.184257602862254e-06,
+      "loss": 4.2409,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2021466905187835e-06,
+      "loss": 4.272,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2200357781753134e-06,
+      "loss": 4.1107,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.237924865831843e-06,
+      "loss": 4.1501,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2558139534883724e-06,
+      "loss": 4.3055,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2737030411449015e-06,
+      "loss": 4.3098,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2915921288014314e-06,
+      "loss": 4.365,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.309481216457961e-06,
+      "loss": 4.176,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3273703041144904e-06,
+      "loss": 4.2073,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3452593917710204e-06,
+      "loss": 4.075,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3631484794275494e-06,
+      "loss": 4.024,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.381037567084079e-06,
+      "loss": 4.3253,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3989266547406084e-06,
+      "loss": 4.1299,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4168157423971384e-06,
+      "loss": 4.0563,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4347048300536674e-06,
+      "loss": 4.1021,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.452593917710197e-06,
+      "loss": 4.0154,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4704830053667264e-06,
+      "loss": 4.1376,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4883720930232564e-06,
+      "loss": 3.8639,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.506261180679786e-06,
+      "loss": 4.0243,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.524150268336315e-06,
+      "loss": 4.2564,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5420393559928444e-06,
+      "loss": 4.1875,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5599284436493744e-06,
+      "loss": 4.2896,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.577817531305904e-06,
+      "loss": 4.1104,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5957066189624334e-06,
+      "loss": 4.3469,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6135957066189624e-06,
+      "loss": 4.1174,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6314847942754924e-06,
+      "loss": 3.7377,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.649373881932022e-06,
+      "loss": 4.0364,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6672629695885514e-06,
+      "loss": 3.8891,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.685152057245081e-06,
+      "loss": 3.7076,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70304114490161e-06,
+      "loss": 3.9099,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72093023255814e-06,
+      "loss": 3.9484,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7388193202146694e-06,
+      "loss": 4.1033,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.756708407871199e-06,
+      "loss": 3.9562,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.774597495527728e-06,
+      "loss": 4.0688,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.792486583184258e-06,
+      "loss": 3.8831,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8103756708407874e-06,
+      "loss": 4.0032,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828264758497317e-06,
+      "loss": 4.1382,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 3.8982,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8640429338103754e-06,
+      "loss": 3.8862,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.881932021466906e-06,
+      "loss": 3.8142,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.899821109123435e-06,
+      "loss": 3.8258,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.917710196779965e-06,
+      "loss": 3.8321,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.935599284436494e-06,
+      "loss": 3.8611,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.953488372093024e-06,
+      "loss": 3.5592,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.971377459749553e-06,
+      "loss": 4.1332,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.989266547406083e-06,
+      "loss": 3.8436,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.007155635062612e-06,
+      "loss": 4.06,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.025044722719142e-06,
+      "loss": 3.922,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.042933810375671e-06,
+      "loss": 3.663,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.060822898032201e-06,
+      "loss": 3.8817,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.07871198568873e-06,
+      "loss": 4.1409,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.09660107334526e-06,
+      "loss": 3.9046,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.114490161001789e-06,
+      "loss": 3.9627,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.132379248658319e-06,
+      "loss": 3.8451,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.150268336314848e-06,
+      "loss": 3.9111,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.168157423971378e-06,
+      "loss": 4.0844,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.186046511627907e-06,
+      "loss": 3.8503,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.203935599284437e-06,
+      "loss": 3.7239,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.221824686940966e-06,
+      "loss": 3.92,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.239713774597496e-06,
+      "loss": 4.0193,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.257602862254025e-06,
+      "loss": 3.5982,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.275491949910555e-06,
+      "loss": 3.8018,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.293381037567084e-06,
+      "loss": 3.8302,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.311270125223614e-06,
+      "loss": 3.8525,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.329159212880143e-06,
+      "loss": 4.1324,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347048300536673e-06,
+      "loss": 3.7559,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.364937388193202e-06,
+      "loss": 3.9672,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.382826475849732e-06,
+      "loss": 3.9315,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.400715563506261e-06,
+      "loss": 3.6718,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.418604651162791e-06,
+      "loss": 3.7502,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.436493738819321e-06,
+      "loss": 3.9674,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.45438282647585e-06,
+      "loss": 3.8471,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.472271914132379e-06,
+      "loss": 4.1864,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.490161001788909e-06,
+      "loss": 3.8636,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.508050089445439e-06,
+      "loss": 3.9458,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.525939177101969e-06,
+      "loss": 3.9614,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.543828264758497e-06,
+      "loss": 3.9395,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.561717352415027e-06,
+      "loss": 3.7918,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.579606440071557e-06,
+      "loss": 3.813,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.597495527728087e-06,
+      "loss": 3.9564,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 3.6088,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.633273703041145e-06,
+      "loss": 3.8454,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.651162790697675e-06,
+      "loss": 3.9436,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.669051878354205e-06,
+      "loss": 3.7143,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.686940966010734e-06,
+      "loss": 3.7048,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704830053667263e-06,
+      "loss": 3.4491,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722719141323792e-06,
+      "loss": 3.5166,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.740608228980323e-06,
+      "loss": 3.7223,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.758497316636852e-06,
+      "loss": 3.8229,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.776386404293382e-06,
+      "loss": 3.7892,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.79427549194991e-06,
+      "loss": 3.4045,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.812164579606441e-06,
+      "loss": 4.0129,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.83005366726297e-06,
+      "loss": 3.5848,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8479427549195e-06,
+      "loss": 3.7753,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.865831842576029e-06,
+      "loss": 3.8695,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.883720930232559e-06,
+      "loss": 3.6897,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.901610017889088e-06,
+      "loss": 3.9062,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.919499105545618e-06,
+      "loss": 3.8658,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937388193202147e-06,
+      "loss": 3.712,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955277280858677e-06,
+      "loss": 3.8967,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973166368515206e-06,
+      "loss": 3.6393,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991055456171736e-06,
+      "loss": 3.7113,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.008944543828264e-06,
+      "loss": 3.6311,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.026833631484795e-06,
+      "loss": 3.6932,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.044722719141324e-06,
+      "loss": 4.0086,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.062611806797854e-06,
+      "loss": 3.8255,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.080500894454383e-06,
+      "loss": 3.6998,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.098389982110913e-06,
+      "loss": 3.6154,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.116279069767442e-06,
+      "loss": 3.5902,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1341681574239725e-06,
+      "loss": 3.8447,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.152057245080502e-06,
+      "loss": 3.4898,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.169946332737031e-06,
+      "loss": 3.6489,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.18783542039356e-06,
+      "loss": 3.5943,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.20572450805009e-06,
+      "loss": 3.689,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.223613595706619e-06,
+      "loss": 3.7085,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.241502683363149e-06,
+      "loss": 3.4523,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.259391771019678e-06,
+      "loss": 3.6554,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2772808586762085e-06,
+      "loss": 3.5784,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.295169946332738e-06,
+      "loss": 3.5654,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3130590339892675e-06,
+      "loss": 3.7325,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.330948121645797e-06,
+      "loss": 3.3983,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.348837209302326e-06,
+      "loss": 3.6436,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.366726296958855e-06,
+      "loss": 3.7828,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 3.7611,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.402504472271914e-06,
+      "loss": 3.5933,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.420393559928444e-06,
+      "loss": 3.7022,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.438282647584974e-06,
+      "loss": 3.4673,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4561717352415035e-06,
+      "loss": 3.5904,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.474060822898033e-06,
+      "loss": 3.7137,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4919499105545625e-06,
+      "loss": 3.6592,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.509838998211091e-06,
+      "loss": 3.8407,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.527728085867621e-06,
+      "loss": 3.4184,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.54561717352415e-06,
+      "loss": 3.462,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.56350626118068e-06,
+      "loss": 3.5885,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.58139534883721e-06,
+      "loss": 3.6429,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5992844364937395e-06,
+      "loss": 3.6981,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.617173524150269e-06,
+      "loss": 3.5655,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6350626118067985e-06,
+      "loss": 3.5486,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.652951699463328e-06,
+      "loss": 3.7819,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6708407871198575e-06,
+      "loss": 3.8035,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.688729874776386e-06,
+      "loss": 3.6721,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.706618962432916e-06,
+      "loss": 3.6661,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.724508050089446e-06,
+      "loss": 3.4551,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7423971377459755e-06,
+      "loss": 3.4324,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.760286225402505e-06,
+      "loss": 3.5175,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7781753130590345e-06,
+      "loss": 3.6472,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.796064400715564e-06,
+      "loss": 3.4361,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8139534883720935e-06,
+      "loss": 3.6593,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.831842576028624e-06,
+      "loss": 3.6661,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.849731663685152e-06,
+      "loss": 3.7119,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.867620751341682e-06,
+      "loss": 3.5348,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8855098389982115e-06,
+      "loss": 3.5357,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.903398926654741e-06,
+      "loss": 3.5394,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9212880143112705e-06,
+      "loss": 3.7379,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9391771019678e-06,
+      "loss": 3.543,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9570661896243295e-06,
+      "loss": 3.5259,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.97495527728086e-06,
+      "loss": 3.8167,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.992844364937389e-06,
+      "loss": 3.2584,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.010733452593919e-06,
+      "loss": 3.6612,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0286225402504475e-06,
+      "loss": 3.4932,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.046511627906977e-06,
+      "loss": 3.4868,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0644007155635065e-06,
+      "loss": 3.6083,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.082289803220036e-06,
+      "loss": 3.7369,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1001788908765655e-06,
+      "loss": 3.4533,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.118067978533095e-06,
+      "loss": 3.5526,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.135957066189625e-06,
+      "loss": 3.4914,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 3.6058,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.171735241502684e-06,
+      "loss": 3.6485,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.189624329159213e-06,
+      "loss": 3.5843,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2075134168157425e-06,
+      "loss": 3.561,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.225402504472272e-06,
+      "loss": 3.4183,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2432915921288015e-06,
+      "loss": 3.4903,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.261180679785331e-06,
+      "loss": 3.6114,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.279069767441861e-06,
+      "loss": 3.6046,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.296958855098391e-06,
+      "loss": 3.6129,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.31484794275492e-06,
+      "loss": 3.605,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.33273703041145e-06,
+      "loss": 3.5619,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.350626118067979e-06,
+      "loss": 3.369,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.368515205724508e-06,
+      "loss": 3.577,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3864042933810375e-06,
+      "loss": 3.5664,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.404293381037567e-06,
+      "loss": 3.4952,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.422182468694097e-06,
+      "loss": 3.4073,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.440071556350627e-06,
+      "loss": 3.37,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.457960644007156e-06,
+      "loss": 3.5803,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.475849731663686e-06,
+      "loss": 3.545,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.493738819320215e-06,
+      "loss": 3.2224,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.511627906976745e-06,
+      "loss": 3.4316,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5295169946332735e-06,
+      "loss": 3.4825,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.547406082289803e-06,
+      "loss": 3.5611,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.565295169946333e-06,
+      "loss": 3.4404,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.583184257602863e-06,
+      "loss": 3.6987,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.601073345259392e-06,
+      "loss": 3.6357,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.618962432915922e-06,
+      "loss": 3.5263,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.636851520572451e-06,
+      "loss": 3.5018,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.654740608228981e-06,
+      "loss": 3.5024,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.67262969588551e-06,
+      "loss": 3.6574,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.690518783542041e-06,
+      "loss": 3.4978,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.7084078711985685e-06,
+      "loss": 3.3556,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.726296958855099e-06,
+      "loss": 3.4278,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.744186046511628e-06,
+      "loss": 3.3233,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.762075134168158e-06,
+      "loss": 3.4356,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.779964221824687e-06,
+      "loss": 3.3444,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.797853309481217e-06,
+      "loss": 3.4266,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.815742397137746e-06,
+      "loss": 3.414,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.833631484794277e-06,
+      "loss": 3.4407,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.851520572450806e-06,
+      "loss": 3.584,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.869409660107335e-06,
+      "loss": 3.5163,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.887298747763864e-06,
+      "loss": 3.4245,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.905187835420394e-06,
+      "loss": 3.5184,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 3.3879,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.940966010733453e-06,
+      "loss": 3.5291,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.958855098389982e-06,
+      "loss": 3.3478,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.976744186046513e-06,
+      "loss": 3.3912,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.994633273703042e-06,
+      "loss": 3.3409,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.012522361359572e-06,
+      "loss": 3.4487,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.030411449016101e-06,
+      "loss": 3.4963,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.04830053667263e-06,
+      "loss": 3.209,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.066189624329159e-06,
+      "loss": 3.1701,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.084078711985689e-06,
+      "loss": 3.4467,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.101967799642218e-06,
+      "loss": 3.273,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.119856887298749e-06,
+      "loss": 3.5041,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.137745974955278e-06,
+      "loss": 3.4005,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.155635062611808e-06,
+      "loss": 3.2033,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.173524150268337e-06,
+      "loss": 3.359,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.191413237924867e-06,
+      "loss": 3.3976,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.209302325581395e-06,
+      "loss": 3.2323,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.227191413237925e-06,
+      "loss": 3.4922,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.245080500894454e-06,
+      "loss": 3.646,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.262969588550985e-06,
+      "loss": 3.3942,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.280858676207514e-06,
+      "loss": 3.3321,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.298747763864044e-06,
+      "loss": 3.4096,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.316636851520573e-06,
+      "loss": 3.3259,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.334525939177103e-06,
+      "loss": 3.18,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.352415026833632e-06,
+      "loss": 3.3779,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.370304114490162e-06,
+      "loss": 3.6173,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.38819320214669e-06,
+      "loss": 3.2123,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.40608228980322e-06,
+      "loss": 3.6875,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.42397137745975e-06,
+      "loss": 3.1774,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.44186046511628e-06,
+      "loss": 3.3412,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.459749552772809e-06,
+      "loss": 3.2291,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.477638640429339e-06,
+      "loss": 3.1264,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.495527728085868e-06,
+      "loss": 3.1393,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.513416815742398e-06,
+      "loss": 3.5423,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.531305903398928e-06,
+      "loss": 3.3433,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.549194991055456e-06,
+      "loss": 3.1141,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.567084078711986e-06,
+      "loss": 3.2197,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.584973166368516e-06,
+      "loss": 3.1819,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.602862254025045e-06,
+      "loss": 3.4312,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.620751341681575e-06,
+      "loss": 3.2909,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.638640429338105e-06,
+      "loss": 3.2628,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.656529516994635e-06,
+      "loss": 3.3693,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.674418604651164e-06,
+      "loss": 3.4708,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 2.9869,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.710196779964223e-06,
+      "loss": 3.3418,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.728085867620751e-06,
+      "loss": 3.3786,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.74597495527728e-06,
+      "loss": 3.1844,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.763864042933812e-06,
+      "loss": 3.1397,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.781753130590341e-06,
+      "loss": 3.2912,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.79964221824687e-06,
+      "loss": 3.2588,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.8175313059034e-06,
+      "loss": 3.3761,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.83542039355993e-06,
+      "loss": 3.3116,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.853309481216459e-06,
+      "loss": 3.1813,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.871198568872989e-06,
+      "loss": 3.1884,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.889087656529516e-06,
+      "loss": 3.0353,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.906976744186048e-06,
+      "loss": 3.201,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.924865831842577e-06,
+      "loss": 3.5307,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.942754919499107e-06,
+      "loss": 3.1738,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.960644007155636e-06,
+      "loss": 2.8024,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.978533094812166e-06,
+      "loss": 3.3323,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.996422182468695e-06,
+      "loss": 3.373,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.014311270125225e-06,
+      "loss": 3.2732,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.032200357781754e-06,
+      "loss": 3.1037,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.050089445438284e-06,
+      "loss": 3.1765,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.067978533094813e-06,
+      "loss": 3.0444,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.085867620751343e-06,
+      "loss": 3.151,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.103756708407872e-06,
+      "loss": 3.3015,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.121645796064402e-06,
+      "loss": 3.1755,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.139534883720931e-06,
+      "loss": 3.1072,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.15742397137746e-06,
+      "loss": 3.1659,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.17531305903399e-06,
+      "loss": 3.1722,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.19320214669052e-06,
+      "loss": 3.0872,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.211091234347049e-06,
+      "loss": 3.2112,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.228980322003579e-06,
+      "loss": 3.3265,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.246869409660108e-06,
+      "loss": 3.3768,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.264758497316638e-06,
+      "loss": 3.0138,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.282647584973167e-06,
+      "loss": 3.4109,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.300536672629697e-06,
+      "loss": 3.0627,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.318425760286226e-06,
+      "loss": 3.0607,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.336314847942756e-06,
+      "loss": 3.0946,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.354203935599285e-06,
+      "loss": 3.0866,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.372093023255815e-06,
+      "loss": 3.1206,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.389982110912344e-06,
+      "loss": 3.23,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.407871198568874e-06,
+      "loss": 3.2486,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.425760286225403e-06,
+      "loss": 3.0711,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.443649373881933e-06,
+      "loss": 3.3111,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 3.1444,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.479427549194992e-06,
+      "loss": 3.1814,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.497316636851521e-06,
+      "loss": 3.2136,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.51520572450805e-06,
+      "loss": 3.1289,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.53309481216458e-06,
+      "loss": 3.2217,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.55098389982111e-06,
+      "loss": 3.354,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.568872987477639e-06,
+      "loss": 3.0317,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.586762075134169e-06,
+      "loss": 3.2301,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.604651162790698e-06,
+      "loss": 3.0376,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.622540250447228e-06,
+      "loss": 3.0859,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.640429338103757e-06,
+      "loss": 3.2219,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.658318425760287e-06,
+      "loss": 2.9363,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.676207513416816e-06,
+      "loss": 3.0631,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.694096601073346e-06,
+      "loss": 3.1712,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.711985688729875e-06,
+      "loss": 3.2138,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.729874776386405e-06,
+      "loss": 2.916,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.747763864042934e-06,
+      "loss": 3.2475,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.765652951699464e-06,
+      "loss": 2.8908,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.783542039355993e-06,
+      "loss": 3.2409,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.801431127012523e-06,
+      "loss": 3.1573,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.819320214669052e-06,
+      "loss": 3.0432,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.837209302325582e-06,
+      "loss": 2.9602,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.855098389982111e-06,
+      "loss": 3.1708,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.872987477638642e-06,
+      "loss": 2.9973,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.890876565295172e-06,
+      "loss": 3.0368,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.9087656529517e-06,
+      "loss": 3.2507,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.926654740608229e-06,
+      "loss": 2.8964,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.944543828264759e-06,
+      "loss": 3.0531,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.962432915921288e-06,
+      "loss": 3.1694,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.980322003577818e-06,
+      "loss": 3.1252,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.998211091234347e-06,
+      "loss": 3.2322,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.016100178890878e-06,
+      "loss": 3.2168,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.033989266547408e-06,
+      "loss": 3.1687,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.051878354203937e-06,
+      "loss": 3.1783,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.069767441860465e-06,
+      "loss": 3.0206,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.087656529516995e-06,
+      "loss": 3.0035,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.105545617173524e-06,
+      "loss": 2.9909,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.123434704830054e-06,
+      "loss": 3.0701,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.141323792486583e-06,
+      "loss": 3.1052,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.159212880143114e-06,
+      "loss": 3.0814,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.177101967799644e-06,
+      "loss": 3.2185,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.194991055456173e-06,
+      "loss": 3.0994,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.212880143112703e-06,
+      "loss": 2.9292,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 2.8589,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.24865831842576e-06,
+      "loss": 2.9239,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.26654740608229e-06,
+      "loss": 2.8274,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.284436493738819e-06,
+      "loss": 2.9619,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.30232558139535e-06,
+      "loss": 3.0478,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.32021466905188e-06,
+      "loss": 3.1,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.33810375670841e-06,
+      "loss": 2.9372,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.355992844364939e-06,
+      "loss": 3.0556,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.373881932021468e-06,
+      "loss": 3.0255,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.391771019677998e-06,
+      "loss": 3.2373,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.409660107334526e-06,
+      "loss": 3.0709,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.427549194991055e-06,
+      "loss": 2.9517,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.445438282647585e-06,
+      "loss": 3.315,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.463327370304116e-06,
+      "loss": 2.9624,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.481216457960645e-06,
+      "loss": 2.9657,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.499105545617175e-06,
+      "loss": 3.0665,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.516994633273704e-06,
+      "loss": 3.0433,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.534883720930234e-06,
+      "loss": 2.8903,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.552772808586763e-06,
+      "loss": 3.191,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.570661896243293e-06,
+      "loss": 3.0972,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.58855098389982e-06,
+      "loss": 3.0421,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.606440071556352e-06,
+      "loss": 3.0147,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.624329159212881e-06,
+      "loss": 3.0274,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.64221824686941e-06,
+      "loss": 3.0846,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.66010733452594e-06,
+      "loss": 2.8545,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.67799642218247e-06,
+      "loss": 3.2473,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.695885509839e-06,
+      "loss": 3.0223,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.713774597495529e-06,
+      "loss": 3.2075,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.731663685152058e-06,
+      "loss": 3.1132,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.749552772808588e-06,
+      "loss": 2.9775,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.767441860465117e-06,
+      "loss": 2.9731,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.785330948121647e-06,
+      "loss": 2.8498,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.803220035778176e-06,
+      "loss": 3.1439,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.821109123434706e-06,
+      "loss": 3.0258,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.838998211091235e-06,
+      "loss": 3.0431,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.856887298747765e-06,
+      "loss": 2.9389,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.874776386404294e-06,
+      "loss": 3.2226,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.892665474060824e-06,
+      "loss": 3.0609,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910554561717353e-06,
+      "loss": 3.2486,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928443649373883e-06,
+      "loss": 3.0045,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.946332737030412e-06,
+      "loss": 3.1773,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.964221824686942e-06,
+      "loss": 3.0568,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982110912343471e-06,
+      "loss": 2.9095,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 2.9711,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999924233392e-06,
+      "loss": 3.1558,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999696933569e-06,
+      "loss": 2.8653,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999931810054e-06,
+      "loss": 3.1167,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998787734312e-06,
+      "loss": 2.8954,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998105834905e-06,
+      "loss": 2.9239,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999727240234e-06,
+      "loss": 3.0405,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999628743664e-06,
+      "loss": 2.807,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999995150937836e-06,
+      "loss": 3.0589,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999993862905962e-06,
+      "loss": 2.9666,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999992423341057e-06,
+      "loss": 2.9103,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999990832243164e-06,
+      "loss": 3.0016,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989089612334e-06,
+      "loss": 3.2164,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999987195448618e-06,
+      "loss": 2.8934,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999985149752072e-06,
+      "loss": 3.1395,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999982952522759e-06,
+      "loss": 3.0583,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999980603760747e-06,
+      "loss": 2.9177,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999978103466105e-06,
+      "loss": 2.9346,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999975451638912e-06,
+      "loss": 2.9892,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999972648279244e-06,
+      "loss": 2.8681,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999969693387189e-06,
+      "loss": 2.9443,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999966586962836e-06,
+      "loss": 3.0767,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999963329006278e-06,
+      "loss": 2.8326,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999959919517616e-06,
+      "loss": 2.9081,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999956358496951e-06,
+      "loss": 3.1035,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999952645944391e-06,
+      "loss": 2.8518,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999948781860052e-06,
+      "loss": 3.0749,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999944766244045e-06,
+      "loss": 2.823,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999940599096498e-06,
+      "loss": 3.0366,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999936280417532e-06,
+      "loss": 2.8098,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999931810207282e-06,
+      "loss": 2.9197,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999927188465884e-06,
+      "loss": 2.9784,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999922415193473e-06,
+      "loss": 2.9275,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999917490390196e-06,
+      "loss": 3.0199,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999912414056205e-06,
+      "loss": 2.8965,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999907186191652e-06,
+      "loss": 2.9759,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999901806796692e-06,
+      "loss": 3.0521,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999896275871492e-06,
+      "loss": 2.8477,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99989059341622e-06,
+      "loss": 2.8475,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999884759431048e-06,
+      "loss": 2.9129,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999878773916149e-06,
+      "loss": 2.8932,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99987263687171e-06,
+      "loss": 3.3522,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999866348297913e-06,
+      "loss": 3.0365,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99985990819495e-06,
+      "loss": 2.9765,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999853316563015e-06,
+      "loss": 3.1116,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999846573402309e-06,
+      "loss": 2.9935,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999839678713037e-06,
+      "loss": 2.9043,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999832632495407e-06,
+      "loss": 2.9094,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999825434749632e-06,
+      "loss": 2.9268,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99981808547593e-06,
+      "loss": 3.0836,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999810584674525e-06,
+      "loss": 3.0335,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999802932345646e-06,
+      "loss": 3.0492,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99979512848952e-06,
+      "loss": 3.1439,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999787173106387e-06,
+      "loss": 3.1838,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999779066196487e-06,
+      "loss": 3.0088,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999770807760066e-06,
+      "loss": 3.0319,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999762397797375e-06,
+      "loss": 2.7964,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999753836308668e-06,
+      "loss": 3.0405,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999745123294204e-06,
+      "loss": 3.0631,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999736258754249e-06,
+      "loss": 3.0532,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999727242689069e-06,
+      "loss": 2.7476,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999718075098938e-06,
+      "loss": 2.9802,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999708755984136e-06,
+      "loss": 3.1431,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999699285344943e-06,
+      "loss": 2.8557,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99968966318165e-06,
+      "loss": 2.752,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99967988949454e-06,
+      "loss": 2.8404,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999669964283919e-06,
+      "loss": 2.9323,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999659887550082e-06,
+      "loss": 3.0273,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999649659293335e-06,
+      "loss": 3.0824,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999639279513991e-06,
+      "loss": 3.0867,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99962874821236e-06,
+      "loss": 3.1224,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999618065388766e-06,
+      "loss": 3.0742,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99960723104353e-06,
+      "loss": 2.7911,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999596245176979e-06,
+      "loss": 2.8513,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99958510778945e-06,
+      "loss": 2.7631,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999573818881275e-06,
+      "loss": 3.2112,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999562378452802e-06,
+      "loss": 3.1656,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999550786504374e-06,
+      "loss": 2.797,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999539043036344e-06,
+      "loss": 3.0371,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999527148049066e-06,
+      "loss": 2.99,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999515101542903e-06,
+      "loss": 2.8404,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999502903518219e-06,
+      "loss": 2.9569,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999490553975384e-06,
+      "loss": 2.9161,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999478052914771e-06,
+      "loss": 3.1001,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99946540033676e-06,
+      "loss": 2.848,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999452596241735e-06,
+      "loss": 2.911,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999439640630083e-06,
+      "loss": 2.9977,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999426533502195e-06,
+      "loss": 3.0013,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999413274858471e-06,
+      "loss": 2.9946,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999399864699314e-06,
+      "loss": 2.8411,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999386303025125e-06,
+      "loss": 2.7113,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999372589836319e-06,
+      "loss": 2.6944,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999358725133312e-06,
+      "loss": 2.8538,
+      "step": 651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999344708916524e-06,
+      "loss": 3.0567,
+      "step": 652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999330541186377e-06,
+      "loss": 2.8349,
+      "step": 653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999316221943303e-06,
+      "loss": 3.1135,
+      "step": 654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999301751187736e-06,
+      "loss": 2.8661,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999287128920112e-06,
+      "loss": 2.8248,
+      "step": 656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999272355140878e-06,
+      "loss": 3.0499,
+      "step": 657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99925742985048e-06,
+      "loss": 2.6041,
+      "step": 658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999242353049367e-06,
+      "loss": 2.8834,
+      "step": 659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999227124738002e-06,
+      "loss": 2.9495,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999211744916841e-06,
+      "loss": 2.9273,
+      "step": 661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999196213586356e-06,
+      "loss": 3.0471,
+      "step": 662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99918053074701e-06,
+      "loss": 2.6971,
+      "step": 663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999164696399287e-06,
+      "loss": 3.0907,
+      "step": 664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99914871054366e-06,
+      "loss": 2.8288,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999132573180615e-06,
+      "loss": 2.7967,
+      "step": 666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999116284310643e-06,
+      "loss": 2.9788,
+      "step": 667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99909984393424e-06,
+      "loss": 2.9878,
+      "step": 668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999083252051895e-06,
+      "loss": 2.8876,
+      "step": 669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999066508664121e-06,
+      "loss": 2.9459,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999049613771421e-06,
+      "loss": 3.1751,
+      "step": 671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999032567374304e-06,
+      "loss": 2.9792,
+      "step": 672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999015369473295e-06,
+      "loss": 2.9679,
+      "step": 673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998998020068906e-06,
+      "loss": 2.665,
+      "step": 674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998980519161668e-06,
+      "loss": 2.8222,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998962866752111e-06,
+      "loss": 2.9209,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998945062840768e-06,
+      "loss": 2.8053,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998927107428181e-06,
+      "loss": 2.7514,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998909000514892e-06,
+      "loss": 2.7616,
+      "step": 679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998890742101453e-06,
+      "loss": 2.8184,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998872332188412e-06,
+      "loss": 2.9296,
+      "step": 681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998853770776333e-06,
+      "loss": 2.8945,
+      "step": 682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998835057865773e-06,
+      "loss": 3.0408,
+      "step": 683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998816193457304e-06,
+      "loss": 2.993,
+      "step": 684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998797177551494e-06,
+      "loss": 2.8621,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998778010148921e-06,
+      "loss": 2.9046,
+      "step": 686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998758691250166e-06,
+      "loss": 3.0602,
+      "step": 687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998739220855812e-06,
+      "loss": 2.857,
+      "step": 688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998719598966455e-06,
+      "loss": 2.6026,
+      "step": 689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998699825582683e-06,
+      "loss": 3.0018,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998679900705099e-06,
+      "loss": 2.899,
+      "step": 691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998659824334307e-06,
+      "loss": 2.9382,
+      "step": 692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998639596470914e-06,
+      "loss": 2.845,
+      "step": 693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998619217115532e-06,
+      "loss": 3.2101,
+      "step": 694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998598686268783e-06,
+      "loss": 2.7933,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998578003931284e-06,
+      "loss": 2.7259,
+      "step": 696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998557170103664e-06,
+      "loss": 2.8627,
+      "step": 697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998536184786556e-06,
+      "loss": 2.7977,
+      "step": 698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998515047980595e-06,
+      "loss": 2.8545,
+      "step": 699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998493759686419e-06,
+      "loss": 2.9276,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998472319904677e-06,
+      "loss": 3.2278,
+      "step": 701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998450728636016e-06,
+      "loss": 2.9303,
+      "step": 702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998428985881092e-06,
+      "loss": 2.8591,
+      "step": 703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998407091640564e-06,
+      "loss": 2.6388,
+      "step": 704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998385045915094e-06,
+      "loss": 2.9291,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998362848705351e-06,
+      "loss": 3.0053,
+      "step": 706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998340500012009e-06,
+      "loss": 2.95,
+      "step": 707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998317999835743e-06,
+      "loss": 2.8927,
+      "step": 708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998295348177237e-06,
+      "loss": 2.8504,
+      "step": 709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998272545037176e-06,
+      "loss": 2.3371,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998249590416251e-06,
+      "loss": 2.7647,
+      "step": 711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998226484315158e-06,
+      "loss": 2.984,
+      "step": 712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9982032267346e-06,
+      "loss": 3.0769,
+      "step": 713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998179817675278e-06,
+      "loss": 2.8262,
+      "step": 714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998156257137901e-06,
+      "loss": 2.821,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998132545123187e-06,
+      "loss": 2.7013,
+      "step": 716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998108681631853e-06,
+      "loss": 2.8556,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998084666664621e-06,
+      "loss": 2.7728,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99806050022222e-06,
+      "loss": 2.7916,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99803618230538e-06,
+      "loss": 2.958,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998011712914841e-06,
+      "loss": 2.7853,
+      "step": 721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997987092051344e-06,
+      "loss": 3.0234,
+      "step": 722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997962319715634e-06,
+      "loss": 2.624,
+      "step": 723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997937395908464e-06,
+      "loss": 2.7761,
+      "step": 724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997912320630587e-06,
+      "loss": 3.0294,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997887093882763e-06,
+      "loss": 2.5602,
+      "step": 726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997861715665758e-06,
+      "loss": 2.8489,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99783618598034e-06,
+      "loss": 2.9611,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997810504827285e-06,
+      "loss": 2.9445,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997784672207369e-06,
+      "loss": 2.7485,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997758688121373e-06,
+      "loss": 3.0328,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997732552570088e-06,
+      "loss": 2.7577,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997706265554306e-06,
+      "loss": 3.162,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997679827074822e-06,
+      "loss": 2.9739,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99765323713244e-06,
+      "loss": 2.9566,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99762649572796e-06,
+      "loss": 2.701,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997599602862198e-06,
+      "loss": 2.8866,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997572558535966e-06,
+      "loss": 2.9763,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997545362750088e-06,
+      "loss": 3.0086,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997518015505382e-06,
+      "loss": 2.7992,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997490516802682e-06,
+      "loss": 2.9233,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997462866642817e-06,
+      "loss": 2.6869,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997435065026629e-06,
+      "loss": 2.8283,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997407111954958e-06,
+      "loss": 2.7144,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997379007428651e-06,
+      "loss": 3.0556,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997350751448562e-06,
+      "loss": 2.7609,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997322344015547e-06,
+      "loss": 2.8684,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997293785130464e-06,
+      "loss": 3.0405,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997265074794182e-06,
+      "loss": 2.9946,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997236213007568e-06,
+      "loss": 2.853,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9972071997715e-06,
+      "loss": 2.7552,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997178035086857e-06,
+      "loss": 2.7498,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997148718954517e-06,
+      "loss": 2.7104,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997119251375377e-06,
+      "loss": 2.8115,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997089632350325e-06,
+      "loss": 2.7009,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997059861880258e-06,
+      "loss": 2.8698,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997029939966084e-06,
+      "loss": 2.6717,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9969998666087e-06,
+      "loss": 2.7661,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996969641809027e-06,
+      "loss": 2.846,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996939265567977e-06,
+      "loss": 2.8445,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99690873788647e-06,
+      "loss": 2.8077,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996878058765434e-06,
+      "loss": 2.8812,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996847228205795e-06,
+      "loss": 2.723,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996816246208489e-06,
+      "loss": 2.7768,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996785112774456e-06,
+      "loss": 2.8182,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996753827904638e-06,
+      "loss": 3.0098,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996722391599985e-06,
+      "loss": 2.816,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996690803861446e-06,
+      "loss": 2.7366,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996659064689985e-06,
+      "loss": 2.9085,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996627174086556e-06,
+      "loss": 2.7903,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99659513205213e-06,
+      "loss": 2.7803,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996562938587678e-06,
+      "loss": 2.7157,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996530593694175e-06,
+      "loss": 2.8515,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996498097372601e-06,
+      "loss": 2.7423,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996465449623943e-06,
+      "loss": 2.7508,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996432650449186e-06,
+      "loss": 2.8022,
+      "step": 776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996399699849328e-06,
+      "loss": 2.8327,
+      "step": 777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996366597825366e-06,
+      "loss": 2.9499,
+      "step": 778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996333344378302e-06,
+      "loss": 2.7631,
+      "step": 779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996299939509148e-06,
+      "loss": 2.878,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996266383218911e-06,
+      "loss": 2.9006,
+      "step": 781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996232675508613e-06,
+      "loss": 2.7605,
+      "step": 782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996198816379273e-06,
+      "loss": 2.8608,
+      "step": 783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996164805831915e-06,
+      "loss": 2.9941,
+      "step": 784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996130643867575e-06,
+      "loss": 2.8605,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996096330487282e-06,
+      "loss": 3.0732,
+      "step": 786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996061865692081e-06,
+      "loss": 2.6397,
+      "step": 787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996027249483015e-06,
+      "loss": 2.8168,
+      "step": 788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995992481861135e-06,
+      "loss": 2.7058,
+      "step": 789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99595756282749e-06,
+      "loss": 2.7575,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995922492383143e-06,
+      "loss": 2.8752,
+      "step": 791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995887270529153e-06,
+      "loss": 2.7396,
+      "step": 792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995851897266592e-06,
+      "loss": 2.6568,
+      "step": 793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995816372596524e-06,
+      "loss": 2.5624,
+      "step": 794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995780696520036e-06,
+      "loss": 2.6766,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995744869038203e-06,
+      "loss": 2.704,
+      "step": 796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99570889015211e-06,
+      "loss": 3.0335,
+      "step": 797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995672759862852e-06,
+      "loss": 2.9173,
+      "step": 798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995636478171518e-06,
+      "loss": 2.8718,
+      "step": 799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995600045079215e-06,
+      "loss": 3.1703,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99556346058704e-06,
+      "loss": 2.9638,
+      "step": 801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995526724696106e-06,
+      "loss": 2.7039,
+      "step": 802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995489837407527e-06,
+      "loss": 2.7598,
+      "step": 803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995452798722416e-06,
+      "loss": 2.8815,
+      "step": 804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9954156086419e-06,
+      "loss": 2.8065,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995378267167104e-06,
+      "loss": 2.8088,
+      "step": 806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995340774299162e-06,
+      "loss": 2.5443,
+      "step": 807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995303130039207e-06,
+      "loss": 2.8532,
+      "step": 808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995265334388382e-06,
+      "loss": 2.7092,
+      "step": 809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995227387347834e-06,
+      "loss": 2.7312,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99518928891871e-06,
+      "loss": 2.7493,
+      "step": 811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995151039102164e-06,
+      "loss": 2.6775,
+      "step": 812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995112637899359e-06,
+      "loss": 2.5362,
+      "step": 813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995074085311455e-06,
+      "loss": 2.6652,
+      "step": 814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995035381339624e-06,
+      "loss": 2.7876,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994996525985037e-06,
+      "loss": 2.62,
+      "step": 816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994957519248872e-06,
+      "loss": 2.9418,
+      "step": 817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99491836113231e-06,
+      "loss": 2.8332,
+      "step": 818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994879051636539e-06,
+      "loss": 2.8473,
+      "step": 819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99483959076275e-06,
+      "loss": 2.7193,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994799978512137e-06,
+      "loss": 2.628,
+      "step": 821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994760214885906e-06,
+      "loss": 2.6931,
+      "step": 822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994720299885256e-06,
+      "loss": 2.6096,
+      "step": 823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9946802335114e-06,
+      "loss": 2.7054,
+      "step": 824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994640015765552e-06,
+      "loss": 2.9626,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99459964664893e-06,
+      "loss": 2.8431,
+      "step": 826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994559126162758e-06,
+      "loss": 2.484,
+      "step": 827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994518454308262e-06,
+      "loss": 2.8368,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994477631086679e-06,
+      "loss": 2.9244,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994436656499242e-06,
+      "loss": 2.7659,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994395530547196e-06,
+      "loss": 2.8504,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994354253231783e-06,
+      "loss": 2.7107,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994312824554261e-06,
+      "loss": 2.9119,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994271244515878e-06,
+      "loss": 2.7381,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9942295131179e-06,
+      "loss": 2.7609,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994187630361588e-06,
+      "loss": 2.9615,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994145596248214e-06,
+      "loss": 2.5085,
+      "step": 837
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99410341077905e-06,
+      "loss": 2.9277,
+      "step": 838
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994061073955375e-06,
+      "loss": 2.7533,
+      "step": 839
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994018585778471e-06,
+      "loss": 2.7877,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993975946249628e-06,
+      "loss": 2.7709,
+      "step": 841
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993933155370137e-06,
+      "loss": 2.7444,
+      "step": 842
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993890213141295e-06,
+      "loss": 2.6242,
+      "step": 843
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993847119564402e-06,
+      "loss": 2.7601,
+      "step": 844
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993803874640765e-06,
+      "loss": 2.862,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993760478371697e-06,
+      "loss": 2.693,
+      "step": 846
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993716930758513e-06,
+      "loss": 2.7752,
+      "step": 847
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993673231802528e-06,
+      "loss": 2.7371,
+      "step": 848
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99362938150507e-06,
+      "loss": 2.8093,
+      "step": 849
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993585379867466e-06,
+      "loss": 2.8625,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993541226891052e-06,
+      "loss": 2.8191,
+      "step": 851
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993496922577165e-06,
+      "loss": 2.6489,
+      "step": 852
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99345246692715e-06,
+      "loss": 2.693,
+      "step": 853
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993407859942348e-06,
+      "loss": 2.7801,
+      "step": 854
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993363101624117e-06,
+      "loss": 2.9317,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993318191973811e-06,
+      "loss": 2.9497,
+      "step": 856
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99327313099279e-06,
+      "loss": 2.8043,
+      "step": 857
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993227918682424e-06,
+      "loss": 2.6581,
+      "step": 858
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993182555044079e-06,
+      "loss": 2.5802,
+      "step": 859
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993137040079131e-06,
+      "loss": 2.6697,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99309137378896e-06,
+      "loss": 2.9121,
+      "step": 861
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993045556174949e-06,
+      "loss": 2.558,
+      "step": 862
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992999587238487e-06,
+      "loss": 2.6389,
+      "step": 863
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992953466980967e-06,
+      "loss": 2.6507,
+      "step": 864
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992907195403787e-06,
+      "loss": 2.8416,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99286077250835e-06,
+      "loss": 2.8249,
+      "step": 866
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992814198296061e-06,
+      "loss": 2.9362,
+      "step": 867
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992767472768334e-06,
+      "loss": 2.7939,
+      "step": 868
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992720595926584e-06,
+      "loss": 2.9788,
+      "step": 869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99267356777223e-06,
+      "loss": 2.7715,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9926263883067e-06,
+      "loss": 2.7433,
+      "step": 871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992579057531424e-06,
+      "loss": 2.9135,
+      "step": 872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992531575447833e-06,
+      "loss": 2.7766,
+      "step": 873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992483942057369e-06,
+      "loss": 2.4915,
+      "step": 874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992436157361474e-06,
+      "loss": 3.085,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992388221361598e-06,
+      "loss": 2.8594,
+      "step": 876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992340134059192e-06,
+      "loss": 2.7769,
+      "step": 877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992291895455715e-06,
+      "loss": 2.6983,
+      "step": 878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992243505552626e-06,
+      "loss": 2.8112,
+      "step": 879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992194964351396e-06,
+      "loss": 2.5039,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992146271853491e-06,
+      "loss": 2.793,
+      "step": 881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992097428060393e-06,
+      "loss": 2.9235,
+      "step": 882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992048432973576e-06,
+      "loss": 2.4453,
+      "step": 883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991999286594527e-06,
+      "loss": 2.6531,
+      "step": 884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991949988924738e-06,
+      "loss": 2.5187,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9919005399657e-06,
+      "loss": 2.6869,
+      "step": 886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991850939718912e-06,
+      "loss": 2.5316,
+      "step": 887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991801188185878e-06,
+      "loss": 2.7545,
+      "step": 888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991751285368108e-06,
+      "loss": 2.7872,
+      "step": 889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99170123126711e-06,
+      "loss": 2.9848,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991651025884404e-06,
+      "loss": 2.7354,
+      "step": 891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99160066922151e-06,
+      "loss": 2.6453,
+      "step": 892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991550161279955e-06,
+      "loss": 2.5007,
+      "step": 893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99149950206127e-06,
+      "loss": 2.9705,
+      "step": 894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991448691566989e-06,
+      "loss": 2.7754,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991397729798653e-06,
+      "loss": 2.7389,
+      "step": 896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991346616757806e-06,
+      "loss": 2.784,
+      "step": 897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991295352445996e-06,
+      "loss": 2.9015,
+      "step": 898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99124393686478e-06,
+      "loss": 2.5002,
+      "step": 899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991192370015714e-06,
+      "loss": 2.7433,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991140651900359e-06,
+      "loss": 2.534,
+      "step": 901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991088782520286e-06,
+      "loss": 2.6068,
+      "step": 902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991036761877066e-06,
+      "loss": 2.9275,
+      "step": 903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990984589972274e-06,
+      "loss": 2.7446,
+      "step": 904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990932266807491e-06,
+      "loss": 2.7086,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990879792384305e-06,
+      "loss": 2.5972,
+      "step": 906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990827166704305e-06,
+      "loss": 2.7978,
+      "step": 907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990774389769087e-06,
+      "loss": 2.5658,
+      "step": 908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990721461580247e-06,
+      "loss": 2.5433,
+      "step": 909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990668382139393e-06,
+      "loss": 2.87,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990615151448134e-06,
+      "loss": 2.5218,
+      "step": 911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990561769508078e-06,
+      "loss": 2.9299,
+      "step": 912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990508236320848e-06,
+      "loss": 2.5463,
+      "step": 913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990454551888065e-06,
+      "loss": 2.5216,
+      "step": 914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990400716211356e-06,
+      "loss": 2.8749,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990346729292353e-06,
+      "loss": 2.7808,
+      "step": 916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99029259113269e-06,
+      "loss": 2.7091,
+      "step": 917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99023830173401e-06,
+      "loss": 2.5295,
+      "step": 918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990183861097956e-06,
+      "loss": 2.7744,
+      "step": 919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99012926922618e-06,
+      "loss": 2.8669,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990074526120338e-06,
+      "loss": 2.6981,
+      "step": 921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990019631782085e-06,
+      "loss": 2.8418,
+      "step": 922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989964586213087e-06,
+      "loss": 2.6341,
+      "step": 923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989909389415014e-06,
+      "loss": 2.9223,
+      "step": 924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989854041389535e-06,
+      "loss": 2.6414,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989798542138329e-06,
+      "loss": 2.8114,
+      "step": 926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989742891663076e-06,
+      "loss": 2.7087,
+      "step": 927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989687089965468e-06,
+      "loss": 2.6511,
+      "step": 928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989631137047193e-06,
+      "loss": 2.9103,
+      "step": 929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989575032909946e-06,
+      "loss": 2.8364,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989518777555426e-06,
+      "loss": 2.9305,
+      "step": 931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989462370985343e-06,
+      "loss": 2.4659,
+      "step": 932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989405813201403e-06,
+      "loss": 2.7074,
+      "step": 933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989349104205319e-06,
+      "loss": 2.7905,
+      "step": 934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989292243998812e-06,
+      "loss": 2.7933,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989235232583602e-06,
+      "loss": 2.8327,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989178069961422e-06,
+      "loss": 2.674,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989120756134002e-06,
+      "loss": 2.8109,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989063291103079e-06,
+      "loss": 2.6945,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989005674870391e-06,
+      "loss": 2.9666,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98894790743769e-06,
+      "loss": 2.9162,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988889988806723e-06,
+      "loss": 2.5392,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988831918979246e-06,
+      "loss": 3.0931,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98877369795702e-06,
+      "loss": 2.861,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98871532574181e-06,
+      "loss": 2.6328,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98865680233538e-06,
+      "loss": 2.887,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98859812773951e-06,
+      "loss": 2.6852,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988539301955976e-06,
+      "loss": 2.6827,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98848032498656e-06,
+      "loss": 2.8829,
+      "step": 949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988421196833052e-06,
+      "loss": 2.7361,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988361917497238e-06,
+      "loss": 2.9279,
+      "step": 951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988302486980921e-06,
+      "loss": 2.8046,
+      "step": 952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988242905285899e-06,
+      "loss": 2.8199,
+      "step": 953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988183172413979e-06,
+      "loss": 2.5989,
+      "step": 954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98812328836697e-06,
+      "loss": 2.7168,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988063253146686e-06,
+      "loss": 2.8044,
+      "step": 956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98800306675495e-06,
+      "loss": 2.6456,
+      "step": 957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987942729193584e-06,
+      "loss": 2.7871,
+      "step": 958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987882240464416e-06,
+      "loss": 2.8511,
+      "step": 959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98782160056928e-06,
+      "loss": 2.95,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987760809510015e-06,
+      "loss": 2.7448,
+      "step": 961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98769986728846e-06,
+      "loss": 2.5876,
+      "step": 962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987638773906463e-06,
+      "loss": 2.62,
+      "step": 963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98757752936588e-06,
+      "loss": 2.5139,
+      "step": 964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987516133668561e-06,
+      "loss": 2.8447,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98745458681637e-06,
+      "loss": 2.8039,
+      "step": 966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987392888811172e-06,
+      "loss": 2.8904,
+      "step": 967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987331039654837e-06,
+      "loss": 2.5934,
+      "step": 968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987269039349236e-06,
+      "loss": 2.5439,
+      "step": 969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987206887896254e-06,
+      "loss": 2.8063,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98714458529777e-06,
+      "loss": 2.7893,
+      "step": 971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987082131555674e-06,
+      "loss": 2.6707,
+      "step": 972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987019526671859e-06,
+      "loss": 2.6108,
+      "step": 973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986956770648222e-06,
+      "loss": 2.6342,
+      "step": 974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986893863486664e-06,
+      "loss": 2.6894,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986830805189092e-06,
+      "loss": 2.7784,
+      "step": 976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986767595757419e-06,
+      "loss": 2.5684,
+      "step": 977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986704235193558e-06,
+      "loss": 2.9263,
+      "step": 978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98664072349943e-06,
+      "loss": 2.9688,
+      "step": 979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98657706067696e-06,
+      "loss": 2.5342,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986513246728077e-06,
+      "loss": 2.6942,
+      "step": 981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986449281654717e-06,
+      "loss": 2.6444,
+      "step": 982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986385165458817e-06,
+      "loss": 2.7031,
+      "step": 983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986320898142318e-06,
+      "loss": 2.6616,
+      "step": 984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986256479707171e-06,
+      "loss": 2.7442,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986191910155328e-06,
+      "loss": 2.6318,
+      "step": 986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986127189488744e-06,
+      "loss": 2.8842,
+      "step": 987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986062317709382e-06,
+      "loss": 2.6496,
+      "step": 988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985997294819208e-06,
+      "loss": 2.3097,
+      "step": 989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985932120820191e-06,
+      "loss": 3.0662,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98586679571431e-06,
+      "loss": 2.717,
+      "step": 991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985801319503541e-06,
+      "loss": 2.8108,
+      "step": 992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985735692189869e-06,
+      "loss": 2.6126,
+      "step": 993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985669913775283e-06,
+      "loss": 2.5712,
+      "step": 994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985603984261778e-06,
+      "loss": 2.6233,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98553790365135e-06,
+      "loss": 2.6697,
+      "step": 996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985471671946004e-06,
+      "loss": 2.7148,
+      "step": 997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985405289147746e-06,
+      "loss": 2.7999,
+      "step": 998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985338755258587e-06,
+      "loss": 2.6124,
+      "step": 999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985272070280545e-06,
+      "loss": 2.7131,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985205234215643e-06,
+      "loss": 2.7975,
+      "step": 1001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9851382470659e-06,
+      "loss": 2.6099,
+      "step": 1002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985071108833352e-06,
+      "loss": 2.3895,
+      "step": 1003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985003819520031e-06,
+      "loss": 2.515,
+      "step": 1004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984936379127978e-06,
+      "loss": 2.8123,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984868787659235e-06,
+      "loss": 2.7953,
+      "step": 1006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984801045115852e-06,
+      "loss": 3.0201,
+      "step": 1007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98473315149988e-06,
+      "loss": 2.6887,
+      "step": 1008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984665106813381e-06,
+      "loss": 2.8211,
+      "step": 1009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984596911058413e-06,
+      "loss": 2.5848,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984528564237044e-06,
+      "loss": 2.7277,
+      "step": 1011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984460066351346e-06,
+      "loss": 2.5125,
+      "step": 1012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984391417403395e-06,
+      "loss": 2.6771,
+      "step": 1013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98432261739527e-06,
+      "loss": 2.6226,
+      "step": 1014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984253666329058e-06,
+      "loss": 2.7661,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984184564206847e-06,
+      "loss": 2.7565,
+      "step": 1016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984115311030733e-06,
+      "loss": 2.9127,
+      "step": 1017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984045906802812e-06,
+      "loss": 2.5544,
+      "step": 1018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983976351525189e-06,
+      "loss": 2.9917,
+      "step": 1019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983906645199974e-06,
+      "loss": 2.7456,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983836787829277e-06,
+      "loss": 2.8278,
+      "step": 1021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983766779415216e-06,
+      "loss": 2.5921,
+      "step": 1022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983696619959914e-06,
+      "loss": 2.6135,
+      "step": 1023
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983626309465495e-06,
+      "loss": 2.5322,
+      "step": 1024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98355584793409e-06,
+      "loss": 2.7921,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983485235367836e-06,
+      "loss": 2.7056,
+      "step": 1026
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983414471768873e-06,
+      "loss": 2.7569,
+      "step": 1027
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983343557139344e-06,
+      "loss": 2.6562,
+      "step": 1028
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9832724914814e-06,
+      "loss": 2.6965,
+      "step": 1029
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983201274797192e-06,
+      "loss": 2.6343,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983129907088883e-06,
+      "loss": 2.8637,
+      "step": 1031
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98305838835863e-06,
+      "loss": 2.8894,
+      "step": 1032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982986718608606e-06,
+      "loss": 2.7962,
+      "step": 1033
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982914897840979e-06,
+      "loss": 2.5682,
+      "step": 1034
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98284292605793e-06,
+      "loss": 2.8465,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982770803261634e-06,
+      "loss": 2.4819,
+      "step": 1036
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982698529454283e-06,
+      "loss": 2.7126,
+      "step": 1037
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982626104638064e-06,
+      "loss": 2.8275,
+      "step": 1038
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982553528815173e-06,
+      "loss": 2.5034,
+      "step": 1039
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982480801987808e-06,
+      "loss": 2.9804,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982407924158176e-06,
+      "loss": 2.6391,
+      "step": 1041
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982334895328483e-06,
+      "loss": 2.8887,
+      "step": 1042
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982261715500942e-06,
+      "loss": 2.5545,
+      "step": 1043
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982188384677775e-06,
+      "loss": 2.8191,
+      "step": 1044
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9821149028612e-06,
+      "loss": 2.7273,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982041270053446e-06,
+      "loss": 2.4891,
+      "step": 1046
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981967486256743e-06,
+      "loss": 2.549,
+      "step": 1047
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981893551473329e-06,
+      "loss": 2.6058,
+      "step": 1048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981819465705443e-06,
+      "loss": 2.57,
+      "step": 1049
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981745228955333e-06,
+      "loss": 2.8411,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981670841225246e-06,
+      "loss": 2.3694,
+      "step": 1051
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981596302517438e-06,
+      "loss": 2.5742,
+      "step": 1052
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981521612834167e-06,
+      "loss": 2.6263,
+      "step": 1053
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981446772177696e-06,
+      "loss": 2.9358,
+      "step": 1054
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981371780550295e-06,
+      "loss": 2.357,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981296637954238e-06,
+      "loss": 2.8342,
+      "step": 1056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981221344391798e-06,
+      "loss": 2.912,
+      "step": 1057
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981145899865262e-06,
+      "loss": 2.7674,
+      "step": 1058
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98107030437691e-06,
+      "loss": 2.6757,
+      "step": 1059
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980994557929039e-06,
+      "loss": 2.3924,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980918660523941e-06,
+      "loss": 2.8256,
+      "step": 1061
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98084261216392e-06,
+      "loss": 2.774,
+      "step": 1062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980766412851277e-06,
+      "loss": 2.7539,
+      "step": 1063
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980690062588323e-06,
+      "loss": 2.8908,
+      "step": 1064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980613561377372e-06,
+      "loss": 2.5818,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980536909220741e-06,
+      "loss": 2.467,
+      "step": 1066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980460106120756e-06,
+      "loss": 2.6979,
+      "step": 1067
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98038315207974e-06,
+      "loss": 2.8962,
+      "step": 1068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98030604710003e-06,
+      "loss": 2.7027,
+      "step": 1069
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98022879118396e-06,
+      "loss": 2.7655,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980151384333872e-06,
+      "loss": 2.8366,
+      "step": 1071
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980073826552113e-06,
+      "loss": 2.5896,
+      "step": 1072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979996117841033e-06,
+      "loss": 2.6228,
+      "step": 1073
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979918258202986e-06,
+      "loss": 2.7513,
+      "step": 1074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979840247640333e-06,
+      "loss": 2.5455,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979762086155438e-06,
+      "loss": 2.4983,
+      "step": 1076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979683773750668e-06,
+      "loss": 2.8073,
+      "step": 1077
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979605310428399e-06,
+      "loss": 2.6104,
+      "step": 1078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979526696191007e-06,
+      "loss": 2.7621,
+      "step": 1079
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979447931040877e-06,
+      "loss": 2.6312,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979369014980394e-06,
+      "loss": 3.1009,
+      "step": 1081
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979289948011947e-06,
+      "loss": 2.6325,
+      "step": 1082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97921073013794e-06,
+      "loss": 2.8877,
+      "step": 1083
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979131361360766e-06,
+      "loss": 2.618,
+      "step": 1084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979051841682835e-06,
+      "loss": 2.4569,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978972171106556e-06,
+      "loss": 2.3164,
+      "step": 1086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978892349634342e-06,
+      "loss": 2.7168,
+      "step": 1087
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978812377268615e-06,
+      "loss": 2.7635,
+      "step": 1088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978732254011793e-06,
+      "loss": 2.6265,
+      "step": 1089
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978651979866312e-06,
+      "loss": 2.5751,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9785715548346e-06,
+      "loss": 2.6865,
+      "step": 1091
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978490978919095e-06,
+      "loss": 2.6915,
+      "step": 1092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978410252122242e-06,
+      "loss": 2.4584,
+      "step": 1093
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978329374446483e-06,
+      "loss": 2.6087,
+      "step": 1094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978248345894271e-06,
+      "loss": 2.8016,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978167166468061e-06,
+      "loss": 2.6206,
+      "step": 1096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978085836170316e-06,
+      "loss": 2.5587,
+      "step": 1097
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978004355003498e-06,
+      "loss": 2.523,
+      "step": 1098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977922722970076e-06,
+      "loss": 2.6969,
+      "step": 1099
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97784094007253e-06,
+      "loss": 2.6684,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97775900631333e-06,
+      "loss": 2.5037,
+      "step": 1101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977676921694963e-06,
+      "loss": 2.5557,
+      "step": 1102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977594686219918e-06,
+      "loss": 2.3207,
+      "step": 1103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977512299890685e-06,
+      "loss": 2.6064,
+      "step": 1104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977429762709765e-06,
+      "loss": 2.8858,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977347074679655e-06,
+      "loss": 2.6668,
+      "step": 1106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977264235802861e-06,
+      "loss": 2.5805,
+      "step": 1107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977181246081895e-06,
+      "loss": 2.8151,
+      "step": 1108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977098105519273e-06,
+      "loss": 2.7285,
+      "step": 1109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977014814117514e-06,
+      "loss": 2.6294,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976931371879142e-06,
+      "loss": 2.7054,
+      "step": 1111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976847778806684e-06,
+      "loss": 2.5843,
+      "step": 1112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976764034902678e-06,
+      "loss": 2.5302,
+      "step": 1113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976680140169657e-06,
+      "loss": 2.7087,
+      "step": 1114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976596094610168e-06,
+      "loss": 2.5536,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976511898226755e-06,
+      "loss": 2.7413,
+      "step": 1116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976427551021972e-06,
+      "loss": 2.723,
+      "step": 1117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976343052998371e-06,
+      "loss": 2.5756,
+      "step": 1118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976258404158518e-06,
+      "loss": 2.3044,
+      "step": 1119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976173604504977e-06,
+      "loss": 2.5653,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976088654040317e-06,
+      "loss": 2.5261,
+      "step": 1121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976003552767112e-06,
+      "loss": 2.6302,
+      "step": 1122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975918300687943e-06,
+      "loss": 2.5113,
+      "step": 1123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97583289780539e-06,
+      "loss": 2.6975,
+      "step": 1124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975747344122048e-06,
+      "loss": 2.6482,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975661639640503e-06,
+      "loss": 2.6482,
+      "step": 1126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975575784363358e-06,
+      "loss": 2.5117,
+      "step": 1127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97548977829321e-06,
+      "loss": 2.7992,
+      "step": 1128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975403621432668e-06,
+      "loss": 2.8124,
+      "step": 1129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975317313784344e-06,
+      "loss": 2.7781,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97523085535085e-06,
+      "loss": 2.9155,
+      "step": 1131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97514424613481e-06,
+      "loss": 2.8018,
+      "step": 1132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975057486138849e-06,
+      "loss": 2.6087,
+      "step": 1133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974970575365595e-06,
+      "loss": 2.4488,
+      "step": 1134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974883513817682e-06,
+      "loss": 2.6394,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974796301497748e-06,
+      "loss": 2.8425,
+      "step": 1136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974708938408437e-06,
+      "loss": 2.6794,
+      "step": 1137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974621424552395e-06,
+      "loss": 2.6495,
+      "step": 1138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974533759932278e-06,
+      "loss": 2.8468,
+      "step": 1139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97444594455074e-06,
+      "loss": 2.5906,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974357978410442e-06,
+      "loss": 2.5863,
+      "step": 1141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974269861514052e-06,
+      "loss": 2.711,
+      "step": 1142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974181593864238e-06,
+      "loss": 2.5578,
+      "step": 1143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974093175463677e-06,
+      "loss": 2.453,
+      "step": 1144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97400460631505e-06,
+      "loss": 2.6743,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973915886421035e-06,
+      "loss": 2.6711,
+      "step": 1146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973827015784328e-06,
+      "loss": 2.8875,
+      "step": 1147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973737994407619e-06,
+      "loss": 2.8053,
+      "step": 1148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973648822293607e-06,
+      "loss": 2.4753,
+      "step": 1149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973559499444993e-06,
+      "loss": 2.5882,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973470025864486e-06,
+      "loss": 2.639,
+      "step": 1151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973380401554797e-06,
+      "loss": 2.6928,
+      "step": 1152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973290626518642e-06,
+      "loss": 2.7132,
+      "step": 1153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973200700758741e-06,
+      "loss": 2.8379,
+      "step": 1154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973110624277819e-06,
+      "loss": 2.2042,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973020397078609e-06,
+      "loss": 2.4258,
+      "step": 1156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972930019163844e-06,
+      "loss": 2.7501,
+      "step": 1157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97283949053626e-06,
+      "loss": 2.6183,
+      "step": 1158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972748811198605e-06,
+      "loss": 2.654,
+      "step": 1159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972657981153624e-06,
+      "loss": 2.8103,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972567000404072e-06,
+      "loss": 2.3998,
+      "step": 1161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972475868952704e-06,
+      "loss": 2.5443,
+      "step": 1162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972384586802284e-06,
+      "loss": 2.6455,
+      "step": 1163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972293153955579e-06,
+      "loss": 2.8007,
+      "step": 1164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972201570415358e-06,
+      "loss": 2.5093,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972109836184395e-06,
+      "loss": 2.6217,
+      "step": 1166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972017951265476e-06,
+      "loss": 2.6963,
+      "step": 1167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97192591566138e-06,
+      "loss": 2.6289,
+      "step": 1168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9718337293749e-06,
+      "loss": 2.5132,
+      "step": 1169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971741392408826e-06,
+      "loss": 2.5125,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971648904765963e-06,
+      "loss": 2.5402,
+      "step": 1171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971556266449107e-06,
+      "loss": 2.6243,
+      "step": 1172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971463477461068e-06,
+      "loss": 2.807,
+      "step": 1173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97137053780466e-06,
+      "loss": 2.783,
+      "step": 1174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971277447482696e-06,
+      "loss": 2.7062,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971184206498e-06,
+      "loss": 2.5548,
+      "step": 1176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971090814853398e-06,
+      "loss": 2.6849,
+      "step": 1177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97099727255172e-06,
+      "loss": 2.6365,
+      "step": 1178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970903579595799e-06,
+      "loss": 2.7037,
+      "step": 1179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970809735988475e-06,
+      "loss": 2.9455,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970715741732597e-06,
+      "loss": 2.5944,
+      "step": 1181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970621596831007e-06,
+      "loss": 2.4148,
+      "step": 1182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970527301286563e-06,
+      "loss": 2.3808,
+      "step": 1183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97043285510212e-06,
+      "loss": 2.4991,
+      "step": 1184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970338258280541e-06,
+      "loss": 2.7227,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970243510824692e-06,
+      "loss": 2.6122,
+      "step": 1186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970148612737447e-06,
+      "loss": 2.5761,
+      "step": 1187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97005356402168e-06,
+      "loss": 2.7891,
+      "step": 1188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969958364680272e-06,
+      "loss": 2.4801,
+      "step": 1189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969863014716108e-06,
+      "loss": 2.8145,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969767514132079e-06,
+      "loss": 2.6722,
+      "step": 1191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96967186293108e-06,
+      "loss": 2.7882,
+      "step": 1192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969576061116005e-06,
+      "loss": 2.9066,
+      "step": 1193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969480108689762e-06,
+      "loss": 2.5391,
+      "step": 1194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969384005655259e-06,
+      "loss": 2.8623,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969287752015406e-06,
+      "loss": 2.755,
+      "step": 1196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96919134777312e-06,
+      "loss": 2.6531,
+      "step": 1197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969094792931326e-06,
+      "loss": 2.7592,
+      "step": 1198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968998087492949e-06,
+      "loss": 2.4231,
+      "step": 1199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968901231460918e-06,
+      "loss": 2.6573,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968804224838168e-06,
+      "loss": 2.6335,
+      "step": 1201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968707067627643e-06,
+      "loss": 2.4219,
+      "step": 1202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968609759832284e-06,
+      "loss": 2.8021,
+      "step": 1203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96851230145504e-06,
+      "loss": 2.7247,
+      "step": 1204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968414692498868e-06,
+      "loss": 2.9929,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968316932966723e-06,
+      "loss": 2.7913,
+      "step": 1206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968219022861567e-06,
+      "loss": 2.772,
+      "step": 1207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968120962186372e-06,
+      "loss": 2.4864,
+      "step": 1208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968022750944104e-06,
+      "loss": 2.4639,
+      "step": 1209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967924389137745e-06,
+      "loss": 2.7588,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967825876770271e-06,
+      "loss": 2.7169,
+      "step": 1211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967727213844672e-06,
+      "loss": 2.6593,
+      "step": 1212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967628400363933e-06,
+      "loss": 2.6541,
+      "step": 1213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967529436331054e-06,
+      "loss": 2.5671,
+      "step": 1214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967430321749032e-06,
+      "loss": 2.7392,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96733105662087e-06,
+      "loss": 2.6977,
+      "step": 1216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967231640949579e-06,
+      "loss": 2.4554,
+      "step": 1217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967132074738168e-06,
+      "loss": 2.6396,
+      "step": 1218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967032357989658e-06,
+      "loss": 2.8398,
+      "step": 1219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96693249070707e-06,
+      "loss": 2.5028,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96683247289343e-06,
+      "loss": 2.5098,
+      "step": 1221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96673230455177e-06,
+      "loss": 2.8837,
+      "step": 1222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966631985685124e-06,
+      "loss": 2.7309,
+      "step": 1223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966531516296536e-06,
+      "loss": 2.544,
+      "step": 1224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966430896389046e-06,
+      "loss": 2.7046,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966330125965709e-06,
+      "loss": 2.2934,
+      "step": 1226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966229205029574e-06,
+      "loss": 2.7655,
+      "step": 1227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966128133583702e-06,
+      "loss": 2.6662,
+      "step": 1228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966026911631156e-06,
+      "loss": 2.6302,
+      "step": 1229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965925539175005e-06,
+      "loss": 2.5197,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965824016218317e-06,
+      "loss": 2.5126,
+      "step": 1231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965722342764173e-06,
+      "loss": 2.6134,
+      "step": 1232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965620518815654e-06,
+      "loss": 2.7554,
+      "step": 1233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965518544375843e-06,
+      "loss": 2.7842,
+      "step": 1234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965416419447834e-06,
+      "loss": 2.6617,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96531414403472e-06,
+      "loss": 2.5626,
+      "step": 1236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9652117181396e-06,
+      "loss": 2.473,
+      "step": 1237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96510914176558e-06,
+      "loss": 2.6653,
+      "step": 1238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965006414915769e-06,
+      "loss": 2.5877,
+      "step": 1239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964903537593278e-06,
+      "loss": 2.5757,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964800509801226e-06,
+      "loss": 2.5078,
+      "step": 1241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964697331542739e-06,
+      "loss": 2.3734,
+      "step": 1242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964594002820938e-06,
+      "loss": 2.5908,
+      "step": 1243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964490523638956e-06,
+      "loss": 2.7969,
+      "step": 1244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96438689399993e-06,
+      "loss": 2.492,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964283113907006e-06,
+      "loss": 2.7731,
+      "step": 1246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964179183363318e-06,
+      "loss": 2.7612,
+      "step": 1247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964075102372026e-06,
+      "loss": 2.7607,
+      "step": 1248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963970870936279e-06,
+      "loss": 2.7155,
+      "step": 1249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963866489059238e-06,
+      "loss": 2.6359,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963761956744064e-06,
+      "loss": 2.68,
+      "step": 1251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963657273993928e-06,
+      "loss": 2.9065,
+      "step": 1252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963552440812e-06,
+      "loss": 2.5012,
+      "step": 1253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963447457201458e-06,
+      "loss": 2.6371,
+      "step": 1254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963342323165485e-06,
+      "loss": 2.7227,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963237038707267e-06,
+      "loss": 2.7259,
+      "step": 1256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963131603829994e-06,
+      "loss": 2.6586,
+      "step": 1257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963026018536862e-06,
+      "loss": 2.6613,
+      "step": 1258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96292028283107e-06,
+      "loss": 2.5506,
+      "step": 1259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962814396715823e-06,
+      "loss": 2.3909,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96270836019433e-06,
+      "loss": 2.4261,
+      "step": 1261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962602173269804e-06,
+      "loss": 2.49,
+      "step": 1262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962495835945465e-06,
+      "loss": 2.7177,
+      "step": 1263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962389348224534e-06,
+      "loss": 2.5812,
+      "step": 1264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962282710110241e-06,
+      "loss": 2.7153,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962175921605812e-06,
+      "loss": 2.6293,
+      "step": 1266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96206898271449e-06,
+      "loss": 2.4801,
+      "step": 1267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961961893439511e-06,
+      "loss": 2.3582,
+      "step": 1268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961854653784123e-06,
+      "loss": 2.5088,
+      "step": 1269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961747263751576e-06,
+      "loss": 2.7722,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961639723345125e-06,
+      "loss": 2.561,
+      "step": 1271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961532032568026e-06,
+      "loss": 2.588,
+      "step": 1272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961424191423547e-06,
+      "loss": 2.7431,
+      "step": 1273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961316199914957e-06,
+      "loss": 2.7766,
+      "step": 1274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961208058045524e-06,
+      "loss": 2.6252,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961099765818526e-06,
+      "loss": 2.6081,
+      "step": 1276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960991323237248e-06,
+      "loss": 2.5555,
+      "step": 1277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960882730304977e-06,
+      "loss": 2.805,
+      "step": 1278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960773987025e-06,
+      "loss": 2.4134,
+      "step": 1279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960665093400616e-06,
+      "loss": 2.5963,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960556049435125e-06,
+      "loss": 2.8096,
+      "step": 1281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960446855131832e-06,
+      "loss": 2.737,
+      "step": 1282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960337510494043e-06,
+      "loss": 2.2282,
+      "step": 1283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960228015525074e-06,
+      "loss": 2.7453,
+      "step": 1284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960118370228244e-06,
+      "loss": 2.8964,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960008574606877e-06,
+      "loss": 2.898,
+      "step": 1286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959898628664298e-06,
+      "loss": 2.611,
+      "step": 1287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95978853240384e-06,
+      "loss": 2.7975,
+      "step": 1288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959678285828839e-06,
+      "loss": 2.6244,
+      "step": 1289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959567888942638e-06,
+      "loss": 2.7095,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959457341748583e-06,
+      "loss": 2.7735,
+      "step": 1291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959346644250021e-06,
+      "loss": 2.5078,
+      "step": 1292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95923579645031e-06,
+      "loss": 2.7034,
+      "step": 1293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95912479835281e-06,
+      "loss": 2.4385,
+      "step": 1294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95901364996088e-06,
+      "loss": 2.5711,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958902351277895e-06,
+      "loss": 2.5071,
+      "step": 1296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958790902307225e-06,
+      "loss": 2.3947,
+      "step": 1297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958679303052245e-06,
+      "loss": 2.5021,
+      "step": 1298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958567553516342e-06,
+      "loss": 2.6165,
+      "step": 1299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958455653702901e-06,
+      "loss": 2.7482,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958343603615311e-06,
+      "loss": 2.339,
+      "step": 1301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95823140325697e-06,
+      "loss": 2.5982,
+      "step": 1302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95811905263128e-06,
+      "loss": 2.5928,
+      "step": 1303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958006551741644e-06,
+      "loss": 2.4336,
+      "step": 1304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95789390059147e-06,
+      "loss": 2.5084,
+      "step": 1305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957781099184175e-06,
+      "loss": 2.8371,
+      "step": 1306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957668147523177e-06,
+      "loss": 2.6218,
+      "step": 1307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957555045611897e-06,
+      "loss": 2.5279,
+      "step": 1308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957441793453764e-06,
+      "loss": 2.6607,
+      "step": 1309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957328391052212e-06,
+      "loss": 2.8073,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957214838410675e-06,
+      "loss": 2.5014,
+      "step": 1311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9571011355326e-06,
+      "loss": 2.3861,
+      "step": 1312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956987282421424e-06,
+      "loss": 2.5139,
+      "step": 1313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956873279080604e-06,
+      "loss": 2.5948,
+      "step": 1314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956759125513593e-06,
+      "loss": 2.4903,
+      "step": 1315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956644821723853e-06,
+      "loss": 2.5712,
+      "step": 1316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956530367714843e-06,
+      "loss": 2.4582,
+      "step": 1317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956415763490038e-06,
+      "loss": 2.5401,
+      "step": 1318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956301009052906e-06,
+      "loss": 2.4266,
+      "step": 1319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956186104406927e-06,
+      "loss": 2.4884,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956071049555585e-06,
+      "loss": 2.3909,
+      "step": 1321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955955844502365e-06,
+      "loss": 2.3851,
+      "step": 1322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955840489250757e-06,
+      "loss": 2.6484,
+      "step": 1323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95572498380426e-06,
+      "loss": 2.4985,
+      "step": 1324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955609328166374e-06,
+      "loss": 2.6162,
+      "step": 1325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955493522340603e-06,
+      "loss": 2.7196,
+      "step": 1326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955377566330455e-06,
+      "loss": 2.5032,
+      "step": 1327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955261460139448e-06,
+      "loss": 2.7251,
+      "step": 1328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955145203771101e-06,
+      "loss": 2.8035,
+      "step": 1329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955028797228933e-06,
+      "loss": 2.6701,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954912240516477e-06,
+      "loss": 2.5446,
+      "step": 1331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95479553363726e-06,
+      "loss": 2.6959,
+      "step": 1332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954678676594824e-06,
+      "loss": 2.416,
+      "step": 1333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954561669392707e-06,
+      "loss": 2.6321,
+      "step": 1334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954444512034459e-06,
+      "loss": 2.4686,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954327204523626e-06,
+      "loss": 2.7011,
+      "step": 1336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954209746863766e-06,
+      "loss": 2.5341,
+      "step": 1337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954092139058437e-06,
+      "loss": 2.543,
+      "step": 1338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953974381111206e-06,
+      "loss": 2.7988,
+      "step": 1339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953856473025638e-06,
+      "loss": 2.5787,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953738414805312e-06,
+      "loss": 2.2429,
+      "step": 1341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9536202064538e-06,
+      "loss": 2.7787,
+      "step": 1342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953501847974689e-06,
+      "loss": 2.5336,
+      "step": 1343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953383339371562e-06,
+      "loss": 2.8126,
+      "step": 1344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953264680648014e-06,
+      "loss": 2.4743,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95314587180764e-06,
+      "loss": 2.7132,
+      "step": 1346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95302691285404e-06,
+      "loss": 2.5753,
+      "step": 1347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952907803790821e-06,
+      "loss": 2.6628,
+      "step": 1348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95278854462159e-06,
+      "loss": 2.4456,
+      "step": 1349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952669135349963e-06,
+      "loss": 2.4678,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95254957597956e-06,
+      "loss": 2.6489,
+      "step": 1351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952429866514004e-06,
+      "loss": 2.4714,
+      "step": 1352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952310006956922e-06,
+      "loss": 3.1034,
+      "step": 1353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952189997311945e-06,
+      "loss": 2.7642,
+      "step": 1354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952069837582713e-06,
+      "loss": 2.655,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951949527772868e-06,
+      "loss": 2.6637,
+      "step": 1356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951829067886053e-06,
+      "loss": 2.4829,
+      "step": 1357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951708457925922e-06,
+      "loss": 2.7178,
+      "step": 1358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951587697896126e-06,
+      "loss": 2.5825,
+      "step": 1359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95146678780033e-06,
+      "loss": 2.6405,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951345727642195e-06,
+      "loss": 2.7683,
+      "step": 1361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951224517425391e-06,
+      "loss": 2.5327,
+      "step": 1362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951103157153592e-06,
+      "loss": 2.5914,
+      "step": 1363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950981646830476e-06,
+      "loss": 2.4526,
+      "step": 1364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950859986459723e-06,
+      "loss": 2.5323,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950738176045024e-06,
+      "loss": 2.4569,
+      "step": 1366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950616215590068e-06,
+      "loss": 2.5862,
+      "step": 1367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950494105098551e-06,
+      "loss": 2.5937,
+      "step": 1368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950371844574176e-06,
+      "loss": 2.7387,
+      "step": 1369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950249434020647e-06,
+      "loss": 2.6028,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950126873441674e-06,
+      "loss": 2.5434,
+      "step": 1371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95000416284097e-06,
+      "loss": 2.6738,
+      "step": 1372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949881302222258e-06,
+      "loss": 2.519,
+      "step": 1373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949758291589255e-06,
+      "loss": 2.6375,
+      "step": 1374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949635130945696e-06,
+      "loss": 2.5837,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949511820295307e-06,
+      "loss": 2.5245,
+      "step": 1376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949388359641831e-06,
+      "loss": 2.5,
+      "step": 1377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949264748989009e-06,
+      "loss": 2.4492,
+      "step": 1378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949140988340583e-06,
+      "loss": 2.7024,
+      "step": 1379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949017077700305e-06,
+      "loss": 2.5327,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948893017071934e-06,
+      "loss": 2.7393,
+      "step": 1381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948768806459227e-06,
+      "loss": 2.9203,
+      "step": 1382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948644445865946e-06,
+      "loss": 2.6674,
+      "step": 1383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948519935295866e-06,
+      "loss": 2.4434,
+      "step": 1384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948395274752754e-06,
+      "loss": 2.4154,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948270464240394e-06,
+      "loss": 2.5332,
+      "step": 1386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948145503762564e-06,
+      "loss": 2.4013,
+      "step": 1387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948020393323054e-06,
+      "loss": 2.6101,
+      "step": 1388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947895132925656e-06,
+      "loss": 2.7295,
+      "step": 1389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947769722574163e-06,
+      "loss": 2.5061,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947644162272378e-06,
+      "loss": 2.677,
+      "step": 1391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947518452024105e-06,
+      "loss": 2.8115,
+      "step": 1392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947392591833156e-06,
+      "loss": 2.6673,
+      "step": 1393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947266581703345e-06,
+      "loss": 2.4963,
+      "step": 1394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947140421638488e-06,
+      "loss": 2.647,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947014111642412e-06,
+      "loss": 2.6927,
+      "step": 1396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946887651718944e-06,
+      "loss": 2.4017,
+      "step": 1397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946761041871914e-06,
+      "loss": 2.6344,
+      "step": 1398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946634282105164e-06,
+      "loss": 2.8432,
+      "step": 1399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946507372422531e-06,
+      "loss": 2.6095,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946380312827865e-06,
+      "loss": 2.5138,
+      "step": 1401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946253103325014e-06,
+      "loss": 2.5635,
+      "step": 1402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946125743917834e-06,
+      "loss": 2.5583,
+      "step": 1403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945998234610185e-06,
+      "loss": 2.5944,
+      "step": 1404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945870575405934e-06,
+      "loss": 2.5674,
+      "step": 1405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945742766308946e-06,
+      "loss": 2.5672,
+      "step": 1406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945614807323094e-06,
+      "loss": 2.6468,
+      "step": 1407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94548669845226e-06,
+      "loss": 2.7426,
+      "step": 1408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945358439700324e-06,
+      "loss": 2.6104,
+      "step": 1409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945230031071172e-06,
+      "loss": 2.4454,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945101472568699e-06,
+      "loss": 2.4298,
+      "step": 1411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944972764196798e-06,
+      "loss": 2.5767,
+      "step": 1412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944843905959372e-06,
+      "loss": 2.664,
+      "step": 1413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944714897860326e-06,
+      "loss": 2.5618,
+      "step": 1414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944585739903568e-06,
+      "loss": 2.6093,
+      "step": 1415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944456432093013e-06,
+      "loss": 2.4915,
+      "step": 1416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944326974432581e-06,
+      "loss": 2.6555,
+      "step": 1417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944197366926195e-06,
+      "loss": 2.5934,
+      "step": 1418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944067609577782e-06,
+      "loss": 2.6809,
+      "step": 1419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943937702391275e-06,
+      "loss": 2.5346,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943807645370613e-06,
+      "loss": 2.55,
+      "step": 1421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943677438519734e-06,
+      "loss": 2.4668,
+      "step": 1422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94354708184259e-06,
+      "loss": 2.5663,
+      "step": 1423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943416575343122e-06,
+      "loss": 2.6504,
+      "step": 1424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943285919025294e-06,
+      "loss": 2.7539,
+      "step": 1425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943155112893061e-06,
+      "loss": 2.6317,
+      "step": 1426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94302415695039e-06,
+      "loss": 2.5331,
+      "step": 1427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94289305120125e-06,
+      "loss": 2.7097,
+      "step": 1428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942761795649612e-06,
+      "loss": 2.455,
+      "step": 1429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942630390299456e-06,
+      "loss": 2.4693,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942498835154762e-06,
+      "loss": 2.5139,
+      "step": 1431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94236713021952e-06,
+      "loss": 2.5145,
+      "step": 1432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942235275497719e-06,
+      "loss": 2.5692,
+      "step": 1433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942103270993356e-06,
+      "loss": 2.6892,
+      "step": 1434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941971116710433e-06,
+      "loss": 2.3483,
+      "step": 1435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941838812652953e-06,
+      "loss": 2.5858,
+      "step": 1436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941706358824927e-06,
+      "loss": 2.4992,
+      "step": 1437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941573755230369e-06,
+      "loss": 2.5316,
+      "step": 1438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941441001873296e-06,
+      "loss": 2.5453,
+      "step": 1439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941308098757734e-06,
+      "loss": 2.7203,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94117504588771e-06,
+      "loss": 2.5838,
+      "step": 1441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941041843267256e-06,
+      "loss": 2.4221,
+      "step": 1442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94090849090041e-06,
+      "loss": 2.3799,
+      "step": 1443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940774988791211e-06,
+      "loss": 2.5205,
+      "step": 1444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940641336943708e-06,
+      "loss": 2.9125,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94050753536195e-06,
+      "loss": 2.6898,
+      "step": 1446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94037358404999e-06,
+      "loss": 2.7089,
+      "step": 1447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940239483011892e-06,
+      "loss": 2.5652,
+      "step": 1448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940105232251716e-06,
+      "loss": 2.5844,
+      "step": 1449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939970831773535e-06,
+      "loss": 2.7968,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939836281581419e-06,
+      "loss": 2.3246,
+      "step": 1451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939701581679445e-06,
+      "loss": 2.593,
+      "step": 1452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939566732071698e-06,
+      "loss": 2.645,
+      "step": 1453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939431732762263e-06,
+      "loss": 2.9108,
+      "step": 1454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939296583755233e-06,
+      "loss": 2.4833,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939161285054704e-06,
+      "loss": 2.7113,
+      "step": 1456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939025836664775e-06,
+      "loss": 2.3872,
+      "step": 1457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93889023858955e-06,
+      "loss": 2.5859,
+      "step": 1458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938754490833141e-06,
+      "loss": 2.6081,
+      "step": 1459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93861859339966e-06,
+      "loss": 2.5991,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938482546293229e-06,
+      "loss": 2.3742,
+      "step": 1461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938346349517966e-06,
+      "loss": 2.3862,
+      "step": 1462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938210003078004e-06,
+      "loss": 2.4459,
+      "step": 1463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938073506977471e-06,
+      "loss": 2.5017,
+      "step": 1464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937936861220505e-06,
+      "loss": 2.7019,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93780006581125e-06,
+      "loss": 2.4748,
+      "step": 1466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937663120753847e-06,
+      "loss": 2.334,
+      "step": 1467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93752602605245e-06,
+      "loss": 2.7123,
+      "step": 1468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937388781711213e-06,
+      "loss": 2.4144,
+      "step": 1469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937251387734293e-06,
+      "loss": 2.4894,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937113844125858e-06,
+      "loss": 2.6091,
+      "step": 1471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936976150890074e-06,
+      "loss": 2.5903,
+      "step": 1472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936838308031116e-06,
+      "loss": 2.5996,
+      "step": 1473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936700315553159e-06,
+      "loss": 2.5796,
+      "step": 1474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936562173460384e-06,
+      "loss": 2.2999,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936423881756983e-06,
+      "loss": 2.6482,
+      "step": 1476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936285440447142e-06,
+      "loss": 2.5592,
+      "step": 1477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93614684953506e-06,
+      "loss": 2.4962,
+      "step": 1478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936008109024936e-06,
+      "loss": 2.7486,
+      "step": 1479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935869218920972e-06,
+      "loss": 2.718,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935730179227383e-06,
+      "loss": 2.3884,
+      "step": 1481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935590989948378e-06,
+      "loss": 2.53,
+      "step": 1482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93545165108818e-06,
+      "loss": 2.5343,
+      "step": 1483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935312162651004e-06,
+      "loss": 2.5736,
+      "step": 1484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935172524641089e-06,
+      "loss": 2.4009,
+      "step": 1485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935032737062656e-06,
+      "loss": 2.7717,
+      "step": 1486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934892799919947e-06,
+      "loss": 2.5624,
+      "step": 1487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9347527132172e-06,
+      "loss": 2.6473,
+      "step": 1488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934612476958665e-06,
+      "loss": 2.5996,
+      "step": 1489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934472091148588e-06,
+      "loss": 2.4574,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934331555791226e-06,
+      "loss": 2.7254,
+      "step": 1491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934190870890838e-06,
+      "loss": 2.6348,
+      "step": 1492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934050036451685e-06,
+      "loss": 2.6706,
+      "step": 1493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93390905247804e-06,
+      "loss": 2.5521,
+      "step": 1494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93376791897417e-06,
+      "loss": 2.4482,
+      "step": 1495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933626635944358e-06,
+      "loss": 2.7251,
+      "step": 1496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933485203392881e-06,
+      "loss": 2.6176,
+      "step": 1497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93334362132403e-06,
+      "loss": 2.5645,
+      "step": 1498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933201889742091e-06,
+      "loss": 2.5323,
+      "step": 1499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933060008651363e-06,
+      "loss": 2.6155,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932917978056145e-06,
+      "loss": 2.5899,
+      "step": 1501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932775797960738e-06,
+      "loss": 2.4326,
+      "step": 1502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932633468369457e-06,
+      "loss": 2.6486,
+      "step": 1503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932490989286614e-06,
+      "loss": 2.3259,
+      "step": 1504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932348360716524e-06,
+      "loss": 2.47,
+      "step": 1505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93220558266351e-06,
+      "loss": 2.3786,
+      "step": 1506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932062655131902e-06,
+      "loss": 2.4812,
+      "step": 1507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93191957812603e-06,
+      "loss": 2.4479,
+      "step": 1508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93177635165023e-06,
+      "loss": 2.2992,
+      "step": 1509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931632975708844e-06,
+      "loss": 2.4846,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931489450306216e-06,
+      "loss": 2.5448,
+      "step": 1511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931345775446695e-06,
+      "loss": 2.6029,
+      "step": 1512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931201951134637e-06,
+      "loss": 2.4258,
+      "step": 1513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9310579773744e-06,
+      "loss": 2.6606,
+      "step": 1514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930913854170347e-06,
+      "loss": 2.8073,
+      "step": 1515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930769581526846e-06,
+      "loss": 2.7506,
+      "step": 1516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930625159448273e-06,
+      "loss": 2.5583,
+      "step": 1517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930480587939e-06,
+      "loss": 2.6441,
+      "step": 1518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930335867003409e-06,
+      "loss": 2.4488,
+      "step": 1519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930190996645887e-06,
+      "loss": 2.5108,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930045976870828e-06,
+      "loss": 2.5868,
+      "step": 1521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929900807682622e-06,
+      "loss": 2.5079,
+      "step": 1522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92975548908567e-06,
+      "loss": 2.904,
+      "step": 1523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929610021084378e-06,
+      "loss": 2.4483,
+      "step": 1524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929464403683152e-06,
+      "loss": 2.6884,
+      "step": 1525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929318636886405e-06,
+      "loss": 2.6243,
+      "step": 1526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929172720698559e-06,
+      "loss": 2.6759,
+      "step": 1527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929026655124032e-06,
+      "loss": 2.48,
+      "step": 1528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928880440167252e-06,
+      "loss": 2.5327,
+      "step": 1529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928734075832649e-06,
+      "loss": 2.4032,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928587562124663e-06,
+      "loss": 2.3072,
+      "step": 1531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92844089904773e-06,
+      "loss": 2.7449,
+      "step": 1532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928294086606296e-06,
+      "loss": 2.4473,
+      "step": 1533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928147124804812e-06,
+      "loss": 2.476,
+      "step": 1534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928000013647729e-06,
+      "loss": 2.6313,
+      "step": 1535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927852753139508e-06,
+      "loss": 2.708,
+      "step": 1536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927705343284613e-06,
+      "loss": 2.5067,
+      "step": 1537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927557784087507e-06,
+      "loss": 2.6439,
+      "step": 1538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927410075552666e-06,
+      "loss": 2.5427,
+      "step": 1539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927262217684564e-06,
+      "loss": 2.3912,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927114210487684e-06,
+      "loss": 2.7373,
+      "step": 1541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92696605396651e-06,
+      "loss": 2.8205,
+      "step": 1542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926817748125536e-06,
+      "loss": 2.4404,
+      "step": 1543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926669292969252e-06,
+      "loss": 2.7139,
+      "step": 1544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926520688502158e-06,
+      "loss": 2.7419,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92637193472876e-06,
+      "loss": 2.5293,
+      "step": 1546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926223031653563e-06,
+      "loss": 2.683,
+      "step": 1547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926073979281081e-06,
+      "loss": 2.2418,
+      "step": 1548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925924777615834e-06,
+      "loss": 2.7964,
+      "step": 1549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925775426662341e-06,
+      "loss": 2.3924,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925625926425129e-06,
+      "loss": 2.391,
+      "step": 1551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925476276908727e-06,
+      "loss": 2.8373,
+      "step": 1552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925326478117673e-06,
+      "loss": 2.6632,
+      "step": 1553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925176530056507e-06,
+      "loss": 2.3329,
+      "step": 1554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925026432729771e-06,
+      "loss": 2.644,
+      "step": 1555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924876186142017e-06,
+      "loss": 2.3762,
+      "step": 1556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924725790297796e-06,
+      "loss": 2.3973,
+      "step": 1557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924575245201667e-06,
+      "loss": 2.6649,
+      "step": 1558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924424550858194e-06,
+      "loss": 2.4592,
+      "step": 1559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92427370727194e-06,
+      "loss": 2.6161,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924122714447479e-06,
+      "loss": 2.6125,
+      "step": 1561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923971572389388e-06,
+      "loss": 2.6222,
+      "step": 1562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923820281102247e-06,
+      "loss": 2.5384,
+      "step": 1563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923668840590642e-06,
+      "loss": 2.1963,
+      "step": 1564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923517250859159e-06,
+      "loss": 2.3987,
+      "step": 1565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923365511912395e-06,
+      "loss": 2.4455,
+      "step": 1566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92321362375495e-06,
+      "loss": 2.4354,
+      "step": 1567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923061586391424e-06,
+      "loss": 2.4147,
+      "step": 1568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922909399826427e-06,
+      "loss": 2.5743,
+      "step": 1569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92275706406457e-06,
+      "loss": 2.2928,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922604579110471e-06,
+      "loss": 2.3515,
+      "step": 1571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92245194496875e-06,
+      "loss": 2.7031,
+      "step": 1572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922299161644033e-06,
+      "loss": 2.1638,
+      "step": 1573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922146229140952e-06,
+      "loss": 2.8379,
+      "step": 1574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92199314746414e-06,
+      "loss": 2.7174,
+      "step": 1575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921839916618237e-06,
+      "loss": 2.6159,
+      "step": 1576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921686536607888e-06,
+      "loss": 2.4971,
+      "step": 1577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92153300743774e-06,
+      "loss": 2.6092,
+      "step": 1578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921379329112446e-06,
+      "loss": 2.5275,
+      "step": 1579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921225501636665e-06,
+      "loss": 2.2112,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921071525015057e-06,
+      "loss": 2.2363,
+      "step": 1581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92091739925229e-06,
+      "loss": 2.5553,
+      "step": 1582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920763124353033e-06,
+      "loss": 2.614,
+      "step": 1583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920608700321965e-06,
+      "loss": 2.7368,
+      "step": 1584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920454127163762e-06,
+      "loss": 2.5129,
+      "step": 1585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920299404883112e-06,
+      "loss": 2.5748,
+      "step": 1586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920144533484702e-06,
+      "loss": 2.6533,
+      "step": 1587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919989512973227e-06,
+      "loss": 2.5507,
+      "step": 1588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919834343353385e-06,
+      "loss": 2.7006,
+      "step": 1589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919679024629875e-06,
+      "loss": 2.712,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919523556807411e-06,
+      "loss": 2.5441,
+      "step": 1591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919367939890701e-06,
+      "loss": 2.5591,
+      "step": 1592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919212173884459e-06,
+      "loss": 2.474,
+      "step": 1593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91905625879341e-06,
+      "loss": 2.4917,
+      "step": 1594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918900194622277e-06,
+      "loss": 2.6545,
+      "step": 1595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91874398137579e-06,
+      "loss": 2.4539,
+      "step": 1596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918587619058681e-06,
+      "loss": 2.7095,
+      "step": 1597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918431107675695e-06,
+      "loss": 2.6055,
+      "step": 1598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91827444723157e-06,
+      "loss": 2.7385,
+      "step": 1599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918117637731056e-06,
+      "loss": 2.6985,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917960679178904e-06,
+      "loss": 2.6065,
+      "step": 1601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91780357157987e-06,
+      "loss": 2.5624,
+      "step": 1602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91764631493872e-06,
+      "loss": 2.5282,
+      "step": 1603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917488909260216e-06,
+      "loss": 2.7206,
+      "step": 1604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917331354549129e-06,
+      "loss": 2.7247,
+      "step": 1605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917173650810235e-06,
+      "loss": 2.5419,
+      "step": 1606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917015798048313e-06,
+      "loss": 2.8038,
+      "step": 1607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916857796268147e-06,
+      "loss": 2.613,
+      "step": 1608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916699645474525e-06,
+      "loss": 2.5664,
+      "step": 1609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91654134567224e-06,
+      "loss": 2.5527,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91638289686609e-06,
+      "loss": 2.7067,
+      "step": 1611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916224299060876e-06,
+      "loss": 2.5545,
+      "step": 1612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916065552261408e-06,
+      "loss": 2.6594,
+      "step": 1613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915906656472493e-06,
+      "loss": 2.4927,
+      "step": 1614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915747611698949e-06,
+      "loss": 2.3812,
+      "step": 1615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915588417945595e-06,
+      "loss": 2.3453,
+      "step": 1616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915429075217255e-06,
+      "loss": 2.3028,
+      "step": 1617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915269583518761e-06,
+      "loss": 2.4574,
+      "step": 1618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915109942854944e-06,
+      "loss": 2.3884,
+      "step": 1619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914950153230641e-06,
+      "loss": 2.5609,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914790214650698e-06,
+      "loss": 2.4971,
+      "step": 1621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914630127119962e-06,
+      "loss": 2.4802,
+      "step": 1622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914469890643282e-06,
+      "loss": 2.5655,
+      "step": 1623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914309505225516e-06,
+      "loss": 2.6165,
+      "step": 1624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914148970871526e-06,
+      "loss": 2.5746,
+      "step": 1625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913988287586174e-06,
+      "loss": 2.5854,
+      "step": 1626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913827455374332e-06,
+      "loss": 2.4314,
+      "step": 1627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913666474240874e-06,
+      "loss": 2.4196,
+      "step": 1628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913505344190679e-06,
+      "loss": 2.4545,
+      "step": 1629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913344065228629e-06,
+      "loss": 2.2885,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913182637359614e-06,
+      "loss": 2.408,
+      "step": 1631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913021060588523e-06,
+      "loss": 2.55,
+      "step": 1632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912859334920257e-06,
+      "loss": 2.468,
+      "step": 1633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912697460359714e-06,
+      "loss": 2.6682,
+      "step": 1634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912535436911803e-06,
+      "loss": 2.2828,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91237326458143e-06,
+      "loss": 2.4332,
+      "step": 1636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912210943373515e-06,
+      "loss": 2.4336,
+      "step": 1637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912048473292972e-06,
+      "loss": 2.5607,
+      "step": 1638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91188585434473e-06,
+      "loss": 2.3182,
+      "step": 1639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911723086533716e-06,
+      "loss": 2.6562,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91156016986486e-06,
+      "loss": 2.2531,
+      "step": 1641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911397104343102e-06,
+      "loss": 2.4423,
+      "step": 1642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911233889973384e-06,
+      "loss": 2.6669,
+      "step": 1643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911070526760652e-06,
+      "loss": 2.4917,
+      "step": 1644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910907014709858e-06,
+      "loss": 2.5018,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910743353825956e-06,
+      "loss": 2.5376,
+      "step": 1646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910579544113905e-06,
+      "loss": 2.4943,
+      "step": 1647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910415585578673e-06,
+      "loss": 2.5146,
+      "step": 1648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910251478225227e-06,
+      "loss": 2.5283,
+      "step": 1649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91008722205854e-06,
+      "loss": 2.4782,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90992281708359e-06,
+      "loss": 2.6063,
+      "step": 1651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909758263305362e-06,
+      "loss": 2.6968,
+      "step": 1652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90959356072884e-06,
+      "loss": 2.4567,
+      "step": 1653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90942870935902e-06,
+      "loss": 2.29,
+      "step": 1654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909263709200892e-06,
+      "loss": 2.688,
+      "step": 1655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909098560259461e-06,
+      "loss": 2.5651,
+      "step": 1656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908933262539728e-06,
+      "loss": 2.5537,
+      "step": 1657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908767816046709e-06,
+      "loss": 2.384,
+      "step": 1658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908602220785413e-06,
+      "loss": 2.4393,
+      "step": 1659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908436476760858e-06,
+      "loss": 2.5905,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908270583978073e-06,
+      "loss": 2.4804,
+      "step": 1661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908104542442082e-06,
+      "loss": 2.5212,
+      "step": 1662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907938352157913e-06,
+      "loss": 2.3232,
+      "step": 1663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90777201313061e-06,
+      "loss": 2.3516,
+      "step": 1664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90760552536521e-06,
+      "loss": 2.5577,
+      "step": 1665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90743888886676e-06,
+      "loss": 2.6058,
+      "step": 1666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907272103640308e-06,
+      "loss": 2.4768,
+      "step": 1667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907105169690914e-06,
+      "loss": 2.3116,
+      "step": 1668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906938087023632e-06,
+      "loss": 2.543,
+      "step": 1669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906770855643526e-06,
+      "loss": 2.5396,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906603475555666e-06,
+      "loss": 2.7788,
+      "step": 1671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906435946765127e-06,
+      "loss": 2.727,
+      "step": 1672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906268269276983e-06,
+      "loss": 2.3615,
+      "step": 1673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906100443096314e-06,
+      "loss": 2.3803,
+      "step": 1674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90593246822821e-06,
+      "loss": 2.5451,
+      "step": 1675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905764344677761e-06,
+      "loss": 2.4205,
+      "step": 1676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905596072450061e-06,
+      "loss": 2.4737,
+      "step": 1677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90542765155021e-06,
+      "loss": 2.3637,
+      "step": 1678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905259081983314e-06,
+      "loss": 2.5286,
+      "step": 1679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90509036375448e-06,
+      "loss": 2.4782,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90492149686882e-06,
+      "loss": 2.5315,
+      "step": 1681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904752481331454e-06,
+      "loss": 2.3146,
+      "step": 1682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904583317147506e-06,
+      "loss": 2.4801,
+      "step": 1683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904414004322099e-06,
+      "loss": 2.3889,
+      "step": 1684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904244542860366e-06,
+      "loss": 2.529,
+      "step": 1685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904074932767443e-06,
+      "loss": 2.7761,
+      "step": 1686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903905174048471e-06,
+      "loss": 2.737,
+      "step": 1687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903735266708592e-06,
+      "loss": 2.3676,
+      "step": 1688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903565210752958e-06,
+      "loss": 2.5796,
+      "step": 1689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903395006186722e-06,
+      "loss": 2.4995,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903224653015042e-06,
+      "loss": 2.6765,
+      "step": 1691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903054151243082e-06,
+      "loss": 2.543,
+      "step": 1692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902883500876008e-06,
+      "loss": 2.5493,
+      "step": 1693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902712701918992e-06,
+      "loss": 2.4391,
+      "step": 1694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90254175437721e-06,
+      "loss": 2.3997,
+      "step": 1695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902370658255846e-06,
+      "loss": 2.5993,
+      "step": 1696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902199413560081e-06,
+      "loss": 2.5693,
+      "step": 1697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902028020295107e-06,
+      "loss": 2.5367,
+      "step": 1698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901856478466119e-06,
+      "loss": 2.3797,
+      "step": 1699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901684788078314e-06,
+      "loss": 2.5404,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901512949136897e-06,
+      "loss": 2.5906,
+      "step": 1701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901340961647075e-06,
+      "loss": 2.6044,
+      "step": 1702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90116882561406e-06,
+      "loss": 2.4303,
+      "step": 1703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90099654104307e-06,
+      "loss": 2.6379,
+      "step": 1704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900824107939324e-06,
+      "loss": 2.5672,
+      "step": 1705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900651526308051e-06,
+      "loss": 2.644,
+      "step": 1706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90047879615448e-06,
+      "loss": 2.5105,
+      "step": 1707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900305917483846e-06,
+      "loss": 2.4189,
+      "step": 1708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900132890301389e-06,
+      "loss": 2.4477,
+      "step": 1709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899959714612351e-06,
+      "loss": 2.5974,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899786390421981e-06,
+      "loss": 2.4361,
+      "step": 1711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899612917735535e-06,
+      "loss": 2.6704,
+      "step": 1712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899439296558265e-06,
+      "loss": 2.2676,
+      "step": 1713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899265526895437e-06,
+      "loss": 2.5633,
+      "step": 1714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899091608752317e-06,
+      "loss": 2.2691,
+      "step": 1715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898917542134172e-06,
+      "loss": 2.6151,
+      "step": 1716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89874332704628e-06,
+      "loss": 2.6121,
+      "step": 1717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898568963493924e-06,
+      "loss": 2.5557,
+      "step": 1718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898394451482383e-06,
+      "loss": 2.4115,
+      "step": 1719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89821979101695e-06,
+      "loss": 2.4024,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898044982102916e-06,
+      "loss": 2.6864,
+      "step": 1721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89787002474558e-06,
+      "loss": 2.602,
+      "step": 1722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897694918950244e-06,
+      "loss": 2.726,
+      "step": 1723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897519664722216e-06,
+      "loss": 2.5193,
+      "step": 1724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897344262066803e-06,
+      "loss": 2.3147,
+      "step": 1725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897168710989327e-06,
+      "loss": 2.3941,
+      "step": 1726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896993011495106e-06,
+      "loss": 2.5678,
+      "step": 1727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896817163589462e-06,
+      "loss": 2.3035,
+      "step": 1728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896641167277728e-06,
+      "loss": 2.5555,
+      "step": 1729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896465022565236e-06,
+      "loss": 2.4264,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896288729457325e-06,
+      "loss": 2.6435,
+      "step": 1731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896112287959336e-06,
+      "loss": 2.6944,
+      "step": 1732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89593569807662e-06,
+      "loss": 2.6219,
+      "step": 1733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895758959814528e-06,
+      "loss": 2.4699,
+      "step": 1734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895582073178412e-06,
+      "loss": 2.5576,
+      "step": 1735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89540503817364e-06,
+      "loss": 2.187,
+      "step": 1736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895227854805571e-06,
+      "loss": 2.3702,
+      "step": 1737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895050523079577e-06,
+      "loss": 2.3873,
+      "step": 1738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894873043001034e-06,
+      "loss": 2.2159,
+      "step": 1739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894695414575317e-06,
+      "loss": 2.5055,
+      "step": 1740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894517637807815e-06,
+      "loss": 2.3329,
+      "step": 1741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894339712703911e-06,
+      "loss": 2.5352,
+      "step": 1742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894161639268997e-06,
+      "loss": 2.4539,
+      "step": 1743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893983417508475e-06,
+      "loss": 2.5564,
+      "step": 1744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893805047427742e-06,
+      "loss": 2.5478,
+      "step": 1745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893626529032204e-06,
+      "loss": 2.4566,
+      "step": 1746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893447862327272e-06,
+      "loss": 2.454,
+      "step": 1747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893269047318361e-06,
+      "loss": 2.3232,
+      "step": 1748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89309008401089e-06,
+      "loss": 2.6598,
+      "step": 1749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892910972410285e-06,
+      "loss": 2.2983,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89273171252197e-06,
+      "loss": 2.3177,
+      "step": 1751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892552304351382e-06,
+      "loss": 2.4565,
+      "step": 1752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892372747903955e-06,
+      "loss": 2.6989,
+      "step": 1753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89219304318513e-06,
+      "loss": 2.5154,
+      "step": 1754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892013190200357e-06,
+      "loss": 2.5405,
+      "step": 1755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891833188955086e-06,
+      "loss": 2.4037,
+      "step": 1756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891653039454771e-06,
+      "loss": 2.4711,
+      "step": 1757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891472741704871e-06,
+      "loss": 2.4283,
+      "step": 1758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891292295710851e-06,
+      "loss": 2.4615,
+      "step": 1759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89111170147818e-06,
+      "loss": 2.587,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890930959012332e-06,
+      "loss": 2.3984,
+      "step": 1761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890750068318782e-06,
+      "loss": 2.4931,
+      "step": 1762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890569029403016e-06,
+      "loss": 2.3868,
+      "step": 1763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890387842270517e-06,
+      "loss": 2.4745,
+      "step": 1764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89020650692678e-06,
+      "loss": 2.6051,
+      "step": 1765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890025023377296e-06,
+      "loss": 2.7431,
+      "step": 1766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88984339162757e-06,
+      "loss": 2.4583,
+      "step": 1767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889661611683101e-06,
+      "loss": 2.3915,
+      "step": 1768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889479683549403e-06,
+      "loss": 2.2596,
+      "step": 1769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88929760723199e-06,
+      "loss": 2.425,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889115382736377e-06,
+      "loss": 2.5019,
+      "step": 1771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888933010068087e-06,
+      "loss": 2.6586,
+      "step": 1772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888750489232648e-06,
+      "loss": 2.4973,
+      "step": 1773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888567820235591e-06,
+      "loss": 2.6549,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888385003082454e-06,
+      "loss": 2.2964,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888202037778776e-06,
+      "loss": 2.4102,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888018924330101e-06,
+      "loss": 2.6827,
+      "step": 1777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887835662741981e-06,
+      "loss": 2.5733,
+      "step": 1778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887652253019969e-06,
+      "loss": 2.5641,
+      "step": 1779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88746869516962e-06,
+      "loss": 2.6513,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887284989196505e-06,
+      "loss": 2.3671,
+      "step": 1781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887101135106183e-06,
+      "loss": 2.472,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886917132904233e-06,
+      "loss": 2.4447,
+      "step": 1783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886732982596226e-06,
+      "loss": 2.4646,
+      "step": 1784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886548684187747e-06,
+      "loss": 2.4933,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88636423768438e-06,
+      "loss": 2.6926,
+      "step": 1786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886179643091713e-06,
+      "loss": 2.653,
+      "step": 1787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885994900415344e-06,
+      "loss": 2.5246,
+      "step": 1788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885810009660868e-06,
+      "loss": 2.5754,
+      "step": 1789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885624970833892e-06,
+      "loss": 2.6691,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885439783940023e-06,
+      "loss": 2.4579,
+      "step": 1791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885254448984871e-06,
+      "loss": 2.5302,
+      "step": 1792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885068965974056e-06,
+      "loss": 2.6344,
+      "step": 1793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8848833349132e-06,
+      "loss": 2.3765,
+      "step": 1794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884697555807925e-06,
+      "loss": 2.4931,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884511628663862e-06,
+      "loss": 2.5333,
+      "step": 1796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884325553486649e-06,
+      "loss": 2.4647,
+      "step": 1797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884139330281921e-06,
+      "loss": 2.4866,
+      "step": 1798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883952959055327e-06,
+      "loss": 2.3571,
+      "step": 1799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883766439812512e-06,
+      "loss": 2.6053,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88357977255913e-06,
+      "loss": 2.8285,
+      "step": 1801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883392957300835e-06,
+      "loss": 2.493,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883205994043292e-06,
+      "loss": 2.5859,
+      "step": 1803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883018882792169e-06,
+      "loss": 2.419,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882831623553133e-06,
+      "loss": 2.2043,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882644216331859e-06,
+      "loss": 2.7203,
+      "step": 1806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882456661134029e-06,
+      "loss": 2.5946,
+      "step": 1807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882268957965326e-06,
+      "loss": 2.4598,
+      "step": 1808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882081106831439e-06,
+      "loss": 2.487,
+      "step": 1809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881893107738061e-06,
+      "loss": 2.5163,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881704960690891e-06,
+      "loss": 3.0472,
+      "step": 1811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881516665695628e-06,
+      "loss": 2.5939,
+      "step": 1812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881328222757983e-06,
+      "loss": 2.5405,
+      "step": 1813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881139631883662e-06,
+      "loss": 2.6199,
+      "step": 1814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880950893078384e-06,
+      "loss": 2.4852,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880762006347868e-06,
+      "loss": 2.555,
+      "step": 1816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880572971697837e-06,
+      "loss": 2.6498,
+      "step": 1817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880383789134023e-06,
+      "loss": 2.4193,
+      "step": 1818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880194458662159e-06,
+      "loss": 2.4779,
+      "step": 1819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88000498028798e-06,
+      "loss": 2.6284,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879815354017233e-06,
+      "loss": 2.6425,
+      "step": 1821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87962557985566e-06,
+      "loss": 2.2489,
+      "step": 1822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879435657809017e-06,
+      "loss": 2.5199,
+      "step": 1823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879245587883058e-06,
+      "loss": 2.5666,
+      "step": 1824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879055370083542e-06,
+      "loss": 2.4037,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878865004416234e-06,
+      "loss": 2.5013,
+      "step": 1826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878674490886904e-06,
+      "loss": 2.2118,
+      "step": 1827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87848382950133e-06,
+      "loss": 2.485,
+      "step": 1828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878293020265281e-06,
+      "loss": 2.7049,
+      "step": 1829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87810206318455e-06,
+      "loss": 2.6612,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877910958264917e-06,
+      "loss": 2.6204,
+      "step": 1831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877719705512177e-06,
+      "loss": 2.88,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877528304932125e-06,
+      "loss": 2.5771,
+      "step": 1833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877336756530563e-06,
+      "loss": 2.6406,
+      "step": 1834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877145060313295e-06,
+      "loss": 2.5751,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876953216286131e-06,
+      "loss": 2.5839,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876761224454886e-06,
+      "loss": 2.3971,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876569084825377e-06,
+      "loss": 2.4856,
+      "step": 1838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876376797403429e-06,
+      "loss": 2.3591,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876184362194868e-06,
+      "loss": 2.766,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875991779205526e-06,
+      "loss": 2.5601,
+      "step": 1841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875799048441242e-06,
+      "loss": 2.6047,
+      "step": 1842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875606169907852e-06,
+      "loss": 2.588,
+      "step": 1843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875413143611208e-06,
+      "loss": 2.6792,
+      "step": 1844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875219969557154e-06,
+      "loss": 2.4661,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875026647751549e-06,
+      "loss": 2.4744,
+      "step": 1846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87483317820025e-06,
+      "loss": 2.7929,
+      "step": 1847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874639560909118e-06,
+      "loss": 2.4883,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874445795884027e-06,
+      "loss": 2.5088,
+      "step": 1849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874251883130843e-06,
+      "loss": 2.5061,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874057822655446e-06,
+      "loss": 2.5364,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873863614463718e-06,
+      "loss": 2.6223,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873669258561542e-06,
+      "loss": 2.444,
+      "step": 1853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873474754954813e-06,
+      "loss": 2.6131,
+      "step": 1854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87328010364942e-06,
+      "loss": 2.4565,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873085304651265e-06,
+      "loss": 2.4261,
+      "step": 1856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872890357966253e-06,
+      "loss": 2.3985,
+      "step": 1857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87269526360029e-06,
+      "loss": 2.3792,
+      "step": 1858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87250002155929e-06,
+      "loss": 2.5503,
+      "step": 1859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872304631849168e-06,
+      "loss": 2.5872,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872109094475849e-06,
+      "loss": 2.5903,
+      "step": 1861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871913409445257e-06,
+      "loss": 2.3862,
+      "step": 1862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871717576763322e-06,
+      "loss": 2.3543,
+      "step": 1863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87152159643598e-06,
+      "loss": 2.5556,
+      "step": 1864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871325468469173e-06,
+      "loss": 2.2726,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87112919286884e-06,
+      "loss": 2.4248,
+      "step": 1866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870932769640932e-06,
+      "loss": 2.411,
+      "step": 1867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870736198791401e-06,
+      "loss": 2.7174,
+      "step": 1868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870539480326207e-06,
+      "loss": 2.3617,
+      "step": 1869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87034261425131e-06,
+      "loss": 2.4705,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870145600572674e-06,
+      "loss": 2.6711,
+      "step": 1871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869948439296274e-06,
+      "loss": 2.5321,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869751130428084e-06,
+      "loss": 2.6885,
+      "step": 1873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869553673974081e-06,
+      "loss": 2.4042,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869356069940255e-06,
+      "loss": 2.5954,
+      "step": 1875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869158318332588e-06,
+      "loss": 2.5118,
+      "step": 1876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868960419157078e-06,
+      "loss": 2.3465,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86876237241972e-06,
+      "loss": 2.7856,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868564178126518e-06,
+      "loss": 2.6479,
+      "step": 1879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868365836283479e-06,
+      "loss": 2.4332,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86816734689661e-06,
+      "loss": 2.5952,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867968709971932e-06,
+      "loss": 2.7413,
+      "step": 1882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867769925515462e-06,
+      "loss": 2.4991,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867570993533224e-06,
+      "loss": 2.7469,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86737191403125e-06,
+      "loss": 2.3881,
+      "step": 1885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86717268701557e-06,
+      "loss": 2.6497,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866973312492221e-06,
+      "loss": 2.4294,
+      "step": 1887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86677379046725e-06,
+      "loss": 2.326,
+      "step": 1888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866574120946701e-06,
+      "loss": 2.5709,
+      "step": 1889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866374303936626e-06,
+      "loss": 2.5093,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866174339443083e-06,
+      "loss": 2.8013,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865974227472127e-06,
+      "loss": 2.4889,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865773968029824e-06,
+      "loss": 2.4791,
+      "step": 1893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865573561122246e-06,
+      "loss": 2.6641,
+      "step": 1894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865373006755467e-06,
+      "loss": 2.4732,
+      "step": 1895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865172304935563e-06,
+      "loss": 2.3169,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864971455668615e-06,
+      "loss": 2.5662,
+      "step": 1897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864770458960715e-06,
+      "loss": 2.3654,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86456931481795e-06,
+      "loss": 2.7813,
+      "step": 1899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864368023246419e-06,
+      "loss": 2.6353,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864166584252219e-06,
+      "loss": 2.5248,
+      "step": 1901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863964997841459e-06,
+      "loss": 2.4558,
+      "step": 1902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863763264020245e-06,
+      "loss": 2.5983,
+      "step": 1903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863561382794694e-06,
+      "loss": 2.8219,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863359354170922e-06,
+      "loss": 2.2706,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863157178155052e-06,
+      "loss": 2.6138,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862954854753213e-06,
+      "loss": 2.6312,
+      "step": 1907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862752383971536e-06,
+      "loss": 2.4666,
+      "step": 1908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862549765816157e-06,
+      "loss": 2.2772,
+      "step": 1909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862347000293215e-06,
+      "loss": 2.4175,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862144087408856e-06,
+      "loss": 2.4992,
+      "step": 1911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86194102716923e-06,
+      "loss": 2.3994,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861737819580493e-06,
+      "loss": 2.2569,
+      "step": 1913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861534464648802e-06,
+      "loss": 2.6928,
+      "step": 1914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86133096238032e-06,
+      "loss": 2.2326,
+      "step": 1915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861127312781214e-06,
+      "loss": 2.5381,
+      "step": 1916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860923515857654e-06,
+      "loss": 2.5953,
+      "step": 1917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860719571615821e-06,
+      "loss": 2.4973,
+      "step": 1918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860515480061894e-06,
+      "loss": 2.6326,
+      "step": 1919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860311241202057e-06,
+      "loss": 2.56,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8601068550425e-06,
+      "loss": 2.4169,
+      "step": 1921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859902321589417e-06,
+      "loss": 2.3598,
+      "step": 1922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859697640849009e-06,
+      "loss": 2.3483,
+      "step": 1923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859492812827477e-06,
+      "loss": 2.5228,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85928783753103e-06,
+      "loss": 2.6256,
+      "step": 1925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85908271496588e-06,
+      "loss": 2.4014,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858877445138243e-06,
+      "loss": 2.0756,
+      "step": 1927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858672028054342e-06,
+      "loss": 2.5312,
+      "step": 1928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858466463720397e-06,
+      "loss": 2.4792,
+      "step": 1929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858260752142644e-06,
+      "loss": 2.576,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858054893327313e-06,
+      "loss": 2.5071,
+      "step": 1931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857848887280648e-06,
+      "loss": 2.26,
+      "step": 1932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857642734008888e-06,
+      "loss": 2.6246,
+      "step": 1933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857436433518281e-06,
+      "loss": 2.2526,
+      "step": 1934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857229985815081e-06,
+      "loss": 2.3541,
+      "step": 1935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857023390905544e-06,
+      "loss": 2.4831,
+      "step": 1936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856816648795932e-06,
+      "loss": 2.5888,
+      "step": 1937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85660975949251e-06,
+      "loss": 2.6582,
+      "step": 1938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856402723001548e-06,
+      "loss": 2.7325,
+      "step": 1939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85619553932932e-06,
+      "loss": 2.491,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855988208482107e-06,
+      "loss": 2.4861,
+      "step": 1941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85578073046619e-06,
+      "loss": 2.562,
+      "step": 1942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85557310528786e-06,
+      "loss": 2.5108,
+      "step": 1943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855365332953406e-06,
+      "loss": 2.4592,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855157413469127e-06,
+      "loss": 2.5589,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854949346841323e-06,
+      "loss": 2.4505,
+      "step": 1946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854741133076302e-06,
+      "loss": 2.2826,
+      "step": 1947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854532772180372e-06,
+      "loss": 2.5801,
+      "step": 1948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85432426415985e-06,
+      "loss": 2.4862,
+      "step": 1949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854115609021052e-06,
+      "loss": 2.5315,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853906806770304e-06,
+      "loss": 2.4312,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853697857413933e-06,
+      "loss": 2.4136,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853488760958274e-06,
+      "loss": 2.5084,
+      "step": 1953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853279517409661e-06,
+      "loss": 2.4467,
+      "step": 1954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853070126774435e-06,
+      "loss": 2.421,
+      "step": 1955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852860589058945e-06,
+      "loss": 2.7127,
+      "step": 1956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852650904269541e-06,
+      "loss": 2.6719,
+      "step": 1957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852441072412575e-06,
+      "loss": 2.745,
+      "step": 1958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85223109349441e-06,
+      "loss": 2.556,
+      "step": 1959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852020967521407e-06,
+      "loss": 2.3132,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851810694499934e-06,
+      "loss": 2.2867,
+      "step": 1961
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851600274436365e-06,
+      "loss": 2.4699,
+      "step": 1962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851389707337077e-06,
+      "loss": 2.1886,
+      "step": 1963
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851178993208452e-06,
+      "loss": 2.4624,
+      "step": 1964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850968132056873e-06,
+      "loss": 2.4555,
+      "step": 1965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850757123888736e-06,
+      "loss": 2.3894,
+      "step": 1966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850545968710433e-06,
+      "loss": 2.3003,
+      "step": 1967
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85033466652836e-06,
+      "loss": 2.6065,
+      "step": 1968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850123217348926e-06,
+      "loss": 2.3714,
+      "step": 1969
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849911621178538e-06,
+      "loss": 2.3022,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849699878023606e-06,
+      "loss": 2.703,
+      "step": 1971
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849487987890553e-06,
+      "loss": 2.626,
+      "step": 1972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849275950785795e-06,
+      "loss": 2.5192,
+      "step": 1973
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84906376671576e-06,
+      "loss": 2.5771,
+      "step": 1974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84885143568688e-06,
+      "loss": 2.5865,
+      "step": 1975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848638957705588e-06,
+      "loss": 2.6159,
+      "step": 1976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848426332778323e-06,
+      "loss": 2.3361,
+      "step": 1977
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848213560911532e-06,
+      "loss": 2.2484,
+      "step": 1978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848000642111662e-06,
+      "loss": 2.4706,
+      "step": 1979
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847787576385165e-06,
+      "loss": 2.5302,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8475743637385e-06,
+      "loss": 2.7043,
+      "step": 1981
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847361004178127e-06,
+      "loss": 2.2556,
+      "step": 1982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847147497710512e-06,
+      "loss": 2.712,
+      "step": 1983
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846933844342126e-06,
+      "loss": 2.2791,
+      "step": 1984
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846720044079445e-06,
+      "loss": 2.5101,
+      "step": 1985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84650609692895e-06,
+      "loss": 2.6699,
+      "step": 1986
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846292002897123e-06,
+      "loss": 2.5609,
+      "step": 1987
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84607776199045e-06,
+      "loss": 2.3609,
+      "step": 1988
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845863374215428e-06,
+      "loss": 2.4501,
+      "step": 1989
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845648839578555e-06,
+      "loss": 2.509,
+      "step": 1990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845434158086328e-06,
+      "loss": 2.7674,
+      "step": 1991
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845219329745256e-06,
+      "loss": 2.786,
+      "step": 1992
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845004354561853e-06,
+      "loss": 2.4251,
+      "step": 1993
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844789232542629e-06,
+      "loss": 2.4912,
+      "step": 1994
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844573963694104e-06,
+      "loss": 2.4241,
+      "step": 1995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844358548022806e-06,
+      "loss": 2.5123,
+      "step": 1996
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844142985535261e-06,
+      "loss": 2.5399,
+      "step": 1997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843927276238001e-06,
+      "loss": 2.4729,
+      "step": 1998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843711420137565e-06,
+      "loss": 2.749,
+      "step": 1999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843495417240496e-06,
+      "loss": 2.0842,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843279267553338e-06,
+      "loss": 2.7226,
+      "step": 2001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843062971082642e-06,
+      "loss": 2.6533,
+      "step": 2002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842846527834963e-06,
+      "loss": 2.613,
+      "step": 2003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842629937816863e-06,
+      "loss": 2.5088,
+      "step": 2004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842413201034904e-06,
+      "loss": 2.3847,
+      "step": 2005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842196317495656e-06,
+      "loss": 2.5598,
+      "step": 2006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84197928720569e-06,
+      "loss": 2.5618,
+      "step": 2007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841762110171584e-06,
+      "loss": 2.4737,
+      "step": 2008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841544786399923e-06,
+      "loss": 2.4632,
+      "step": 2009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84132731589729e-06,
+      "loss": 2.6218,
+      "step": 2010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841109698670274e-06,
+      "loss": 2.4744,
+      "step": 2011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840891934725477e-06,
+      "loss": 2.4778,
+      "step": 2012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840674024069493e-06,
+      "loss": 2.3254,
+      "step": 2013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840455966708927e-06,
+      "loss": 2.5816,
+      "step": 2014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840237762650389e-06,
+      "loss": 2.372,
+      "step": 2015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84001941190049e-06,
+      "loss": 2.6325,
+      "step": 2016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839800914465852e-06,
+      "loss": 2.5665,
+      "step": 2017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83958227035309e-06,
+      "loss": 2.4467,
+      "step": 2018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839363479568836e-06,
+      "loss": 2.4875,
+      "step": 2019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83914454211972e-06,
+      "loss": 2.5015,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838925458012375e-06,
+      "loss": 2.3486,
+      "step": 2021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838706227253443e-06,
+      "loss": 2.5507,
+      "step": 2022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838486849849566e-06,
+      "loss": 2.3828,
+      "step": 2023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838267325807393e-06,
+      "loss": 2.5029,
+      "step": 2024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838047655133579e-06,
+      "loss": 2.4817,
+      "step": 2025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83782783783478e-06,
+      "loss": 2.6619,
+      "step": 2026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837607873917659e-06,
+      "loss": 2.3397,
+      "step": 2027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837387763388879e-06,
+      "loss": 2.4709,
+      "step": 2028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837167506255115e-06,
+      "loss": 2.6939,
+      "step": 2029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836947102523041e-06,
+      "loss": 2.6962,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836726552199336e-06,
+      "loss": 2.5386,
+      "step": 2031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836505855290683e-06,
+      "loss": 2.5539,
+      "step": 2032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836285011803773e-06,
+      "loss": 2.4897,
+      "step": 2033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836064021745298e-06,
+      "loss": 2.4276,
+      "step": 2034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835842885121955e-06,
+      "loss": 2.3038,
+      "step": 2035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835621601940449e-06,
+      "loss": 2.5777,
+      "step": 2036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83540017220748e-06,
+      "loss": 2.3759,
+      "step": 2037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835178595929764e-06,
+      "loss": 2.578,
+      "step": 2038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834956873114016e-06,
+      "loss": 2.6936,
+      "step": 2039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834735003766954e-06,
+      "loss": 2.5656,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8345129878953e-06,
+      "loss": 2.4815,
+      "step": 2041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834290825505787e-06,
+      "loss": 2.5693,
+      "step": 2042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834068516605146e-06,
+      "loss": 2.5685,
+      "step": 2043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833846061200114e-06,
+      "loss": 2.5776,
+      "step": 2044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833623459297434e-06,
+      "loss": 2.5217,
+      "step": 2045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833400710903853e-06,
+      "loss": 2.512,
+      "step": 2046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833177816026118e-06,
+      "loss": 2.4231,
+      "step": 2047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832954774670988e-06,
+      "loss": 2.6888,
+      "step": 2048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832731586845222e-06,
+      "loss": 2.3728,
+      "step": 2049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832508252555582e-06,
+      "loss": 2.4805,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832284771808837e-06,
+      "loss": 2.4244,
+      "step": 2051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832061144611764e-06,
+      "loss": 2.3899,
+      "step": 2052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831837370971136e-06,
+      "loss": 2.5877,
+      "step": 2053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831613450893735e-06,
+      "loss": 2.4822,
+      "step": 2054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831389384386349e-06,
+      "loss": 2.4994,
+      "step": 2055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831165171455768e-06,
+      "loss": 2.6209,
+      "step": 2056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830940812108787e-06,
+      "loss": 2.3883,
+      "step": 2057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830716306352206e-06,
+      "loss": 2.4524,
+      "step": 2058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83049165419283e-06,
+      "loss": 2.4802,
+      "step": 2059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830266855637462e-06,
+      "loss": 2.5234,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830041910692923e-06,
+      "loss": 2.3353,
+      "step": 2061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829816819366025e-06,
+      "loss": 2.7939,
+      "step": 2062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829591581663592e-06,
+      "loss": 2.4235,
+      "step": 2063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829366197592449e-06,
+      "loss": 2.4363,
+      "step": 2064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829140667159428e-06,
+      "loss": 2.6573,
+      "step": 2065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82891499037136e-06,
+      "loss": 2.5759,
+      "step": 2066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82868916723509e-06,
+      "loss": 2.2924,
+      "step": 2067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828463197757461e-06,
+      "loss": 2.5676,
+      "step": 2068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828237081945318e-06,
+      "loss": 2.4894,
+      "step": 2069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828010819805516e-06,
+      "loss": 2.2883,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827784411344912e-06,
+      "loss": 2.3134,
+      "step": 2071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827557856570367e-06,
+      "loss": 2.5497,
+      "step": 2072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827331155488749e-06,
+      "loss": 2.6164,
+      "step": 2073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827104308106925e-06,
+      "loss": 2.3616,
+      "step": 2074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826877314431775e-06,
+      "loss": 2.2517,
+      "step": 2075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826650174470175e-06,
+      "loss": 2.6481,
+      "step": 2076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82642288822901e-06,
+      "loss": 2.8248,
+      "step": 2077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826195455715166e-06,
+      "loss": 2.515,
+      "step": 2078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82596787693554e-06,
+      "loss": 2.3852,
+      "step": 2079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825740151897025e-06,
+      "loss": 2.3267,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825512280606524e-06,
+      "loss": 2.3647,
+      "step": 2081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825284263070944e-06,
+      "loss": 2.3778,
+      "step": 2082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825056099297196e-06,
+      "loss": 2.3772,
+      "step": 2083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824827789292194e-06,
+      "loss": 2.4927,
+      "step": 2084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824599333062855e-06,
+      "loss": 2.734,
+      "step": 2085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824370730616104e-06,
+      "loss": 2.5736,
+      "step": 2086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824141981958872e-06,
+      "loss": 2.6094,
+      "step": 2087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823913087098089e-06,
+      "loss": 2.5215,
+      "step": 2088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823684046040691e-06,
+      "loss": 2.3,
+      "step": 2089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823454858793621e-06,
+      "loss": 2.5092,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823225525363826e-06,
+      "loss": 2.4998,
+      "step": 2091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822996045758256e-06,
+      "loss": 2.4803,
+      "step": 2092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822766419983864e-06,
+      "loss": 2.3902,
+      "step": 2093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82253664804761e-06,
+      "loss": 2.5001,
+      "step": 2094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822306729956459e-06,
+      "loss": 2.3298,
+      "step": 2095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822076665717376e-06,
+      "loss": 2.3167,
+      "step": 2096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821846455337335e-06,
+      "loss": 2.2594,
+      "step": 2097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821616098823315e-06,
+      "loss": 2.4624,
+      "step": 2098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821385596182296e-06,
+      "loss": 2.4424,
+      "step": 2099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82115494742126e-06,
+      "loss": 2.4942,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820924152547205e-06,
+      "loss": 2.6681,
+      "step": 2101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820693211567117e-06,
+      "loss": 2.5437,
+      "step": 2102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820462124488002e-06,
+      "loss": 2.1204,
+      "step": 2103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820230891316858e-06,
+      "loss": 2.6136,
+      "step": 2104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819999512060697e-06,
+      "loss": 2.4944,
+      "step": 2105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819767986726529e-06,
+      "loss": 2.5096,
+      "step": 2106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819536315321374e-06,
+      "loss": 2.3156,
+      "step": 2107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81930449785225e-06,
+      "loss": 2.344,
+      "step": 2108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81907253432618e-06,
+      "loss": 2.332,
+      "step": 2109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8188404247502e-06,
+      "loss": 2.4346,
+      "step": 2110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818608169131342e-06,
+      "loss": 2.4096,
+      "step": 2111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818375767476644e-06,
+      "loss": 2.5005,
+      "step": 2112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818143219793152e-06,
+      "loss": 2.3291,
+      "step": 2113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81791052608791e-06,
+      "loss": 2.5901,
+      "step": 2114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817677686367973e-06,
+      "loss": 2.4485,
+      "step": 2115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817444700640394e-06,
+      "loss": 2.6154,
+      "step": 2116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817211568912239e-06,
+      "loss": 2.4866,
+      "step": 2117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816978291190572e-06,
+      "loss": 2.691,
+      "step": 2118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81674486748246e-06,
+      "loss": 2.5663,
+      "step": 2119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81651129779498e-06,
+      "loss": 2.3115,
+      "step": 2120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816277582135209e-06,
+      "loss": 2.3529,
+      "step": 2121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816043720510231e-06,
+      "loss": 2.432,
+      "step": 2122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815809712927135e-06,
+      "loss": 2.5603,
+      "step": 2123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81557555939301e-06,
+      "loss": 2.5764,
+      "step": 2124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815341259914955e-06,
+      "loss": 2.349,
+      "step": 2125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81510681450007e-06,
+      "loss": 2.5409,
+      "step": 2126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81487222315546e-06,
+      "loss": 2.3479,
+      "step": 2127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814637485888235e-06,
+      "loss": 2.5203,
+      "step": 2128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814402602705507e-06,
+      "loss": 2.3141,
+      "step": 2129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814167573614398e-06,
+      "loss": 2.4811,
+      "step": 2130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813932398622028e-06,
+      "loss": 2.3329,
+      "step": 2131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813697077735526e-06,
+      "loss": 2.4134,
+      "step": 2132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813461610962026e-06,
+      "loss": 2.2998,
+      "step": 2133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813225998308659e-06,
+      "loss": 2.3418,
+      "step": 2134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81299023978257e-06,
+      "loss": 2.5532,
+      "step": 2135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8127543353909e-06,
+      "loss": 2.3783,
+      "step": 2136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812518285140801e-06,
+      "loss": 2.1607,
+      "step": 2137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81228208903943e-06,
+      "loss": 2.1958,
+      "step": 2138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812045747093939e-06,
+      "loss": 2.4146,
+      "step": 2139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811809259311496e-06,
+      "loss": 2.624,
+      "step": 2140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811572625699263e-06,
+      "loss": 2.7203,
+      "step": 2141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811335846264415e-06,
+      "loss": 2.4295,
+      "step": 2142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81109892101413e-06,
+      "loss": 2.4947,
+      "step": 2143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810861849955585e-06,
+      "loss": 2.4774,
+      "step": 2144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810624633095965e-06,
+      "loss": 2.4429,
+      "step": 2145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810387270442458e-06,
+      "loss": 2.6058,
+      "step": 2146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810149762002263e-06,
+      "loss": 2.4799,
+      "step": 2147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809912107782573e-06,
+      "loss": 2.5764,
+      "step": 2148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809674307790593e-06,
+      "loss": 2.9224,
+      "step": 2149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809436362033528e-06,
+      "loss": 2.4148,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809198270518592e-06,
+      "loss": 2.7392,
+      "step": 2151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808960033252998e-06,
+      "loss": 2.297,
+      "step": 2152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808721650243967e-06,
+      "loss": 2.5875,
+      "step": 2153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808483121498726e-06,
+      "loss": 2.622,
+      "step": 2154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8082444470245e-06,
+      "loss": 2.5731,
+      "step": 2155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808005626828525e-06,
+      "loss": 2.2002,
+      "step": 2156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807766660918038e-06,
+      "loss": 2.4838,
+      "step": 2157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807527549300281e-06,
+      "loss": 2.4006,
+      "step": 2158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807288291982503e-06,
+      "loss": 2.6547,
+      "step": 2159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807048888971951e-06,
+      "loss": 2.6357,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806809340275885e-06,
+      "loss": 2.4674,
+      "step": 2161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806569645901561e-06,
+      "loss": 2.3112,
+      "step": 2162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806329805856245e-06,
+      "loss": 2.6533,
+      "step": 2163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806089820147206e-06,
+      "loss": 2.3747,
+      "step": 2164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805849688781718e-06,
+      "loss": 2.2956,
+      "step": 2165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805609411767057e-06,
+      "loss": 2.4805,
+      "step": 2166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805368989110505e-06,
+      "loss": 2.6782,
+      "step": 2167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80512842081935e-06,
+      "loss": 2.4687,
+      "step": 2168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804887706900881e-06,
+      "loss": 2.4058,
+      "step": 2169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804646847362394e-06,
+      "loss": 2.4855,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804405842211186e-06,
+      "loss": 2.2018,
+      "step": 2171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804164691454567e-06,
+      "loss": 2.6424,
+      "step": 2172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803923395099842e-06,
+      "loss": 2.1911,
+      "step": 2173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803681953154323e-06,
+      "loss": 2.4266,
+      "step": 2174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803440365625327e-06,
+      "loss": 2.641,
+      "step": 2175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803198632520179e-06,
+      "loss": 2.338,
+      "step": 2176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802956753846201e-06,
+      "loss": 2.5557,
+      "step": 2177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802714729610727e-06,
+      "loss": 2.6693,
+      "step": 2178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80247255982109e-06,
+      "loss": 2.4439,
+      "step": 2179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802230244484632e-06,
+      "loss": 2.4773,
+      "step": 2180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801987783608694e-06,
+      "loss": 2.4767,
+      "step": 2181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801745177200622e-06,
+      "loss": 2.2506,
+      "step": 2182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801502425267773e-06,
+      "loss": 2.3291,
+      "step": 2183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801259527817504e-06,
+      "loss": 2.3092,
+      "step": 2184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801016484857174e-06,
+      "loss": 2.26,
+      "step": 2185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800773296394149e-06,
+      "loss": 2.3466,
+      "step": 2186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800529962435799e-06,
+      "loss": 2.3722,
+      "step": 2187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800286482989501e-06,
+      "loss": 2.4014,
+      "step": 2188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800042858062633e-06,
+      "loss": 2.2956,
+      "step": 2189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799799087662577e-06,
+      "loss": 2.3799,
+      "step": 2190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799555171796722e-06,
+      "loss": 2.7596,
+      "step": 2191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79931111047246e-06,
+      "loss": 2.6055,
+      "step": 2192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79906690369719e-06,
+      "loss": 2.6812,
+      "step": 2193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798822551478309e-06,
+      "loss": 2.3937,
+      "step": 2194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798578053823224e-06,
+      "loss": 2.3446,
+      "step": 2195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798333410739345e-06,
+      "loss": 2.2336,
+      "step": 2196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798088622234088e-06,
+      "loss": 2.5234,
+      "step": 2197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797843688314869e-06,
+      "loss": 2.5823,
+      "step": 2198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797598608989115e-06,
+      "loss": 2.4564,
+      "step": 2199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79735338426425e-06,
+      "loss": 2.5781,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797108014147707e-06,
+      "loss": 2.6006,
+      "step": 2201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796862498646921e-06,
+      "loss": 2.4557,
+      "step": 2202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796616837769335e-06,
+      "loss": 2.3888,
+      "step": 2203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796371031522391e-06,
+      "loss": 2.3921,
+      "step": 2204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796125079913543e-06,
+      "loss": 2.3981,
+      "step": 2205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795878982950243e-06,
+      "loss": 2.4099,
+      "step": 2206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795632740639948e-06,
+      "loss": 2.5686,
+      "step": 2207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795386352990125e-06,
+      "loss": 2.4551,
+      "step": 2208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795139820008234e-06,
+      "loss": 2.5817,
+      "step": 2209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794893141701752e-06,
+      "loss": 2.3919,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794646318078153e-06,
+      "loss": 2.4158,
+      "step": 2211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79439934914492e-06,
+      "loss": 2.4646,
+      "step": 2212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794152234909533e-06,
+      "loss": 2.4753,
+      "step": 2213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793904975379486e-06,
+      "loss": 2.5453,
+      "step": 2214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793657570562272e-06,
+      "loss": 2.2306,
+      "step": 2215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793410020465386e-06,
+      "loss": 2.4296,
+      "step": 2216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793162325096331e-06,
+      "loss": 2.4514,
+      "step": 2217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792914484462618e-06,
+      "loss": 2.3821,
+      "step": 2218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792666498571752e-06,
+      "loss": 2.5534,
+      "step": 2219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792418367431253e-06,
+      "loss": 2.3623,
+      "step": 2220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792170091048638e-06,
+      "loss": 2.2864,
+      "step": 2221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791921669431435e-06,
+      "loss": 2.439,
+      "step": 2222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791673102587172e-06,
+      "loss": 2.5989,
+      "step": 2223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79142439052338e-06,
+      "loss": 2.4722,
+      "step": 2224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791175533247597e-06,
+      "loss": 2.4908,
+      "step": 2225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790926530767366e-06,
+      "loss": 2.642,
+      "step": 2226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790677383090235e-06,
+      "loss": 2.6283,
+      "step": 2227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790428090223751e-06,
+      "loss": 2.1725,
+      "step": 2228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790178652175474e-06,
+      "loss": 2.5953,
+      "step": 2229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78992906895296e-06,
+      "loss": 2.3144,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789679340563772e-06,
+      "loss": 2.7335,
+      "step": 2231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789429467015484e-06,
+      "loss": 2.2582,
+      "step": 2232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789179448315662e-06,
+      "loss": 2.6065,
+      "step": 2233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788929284471888e-06,
+      "loss": 2.5658,
+      "step": 2234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788678975491741e-06,
+      "loss": 2.3066,
+      "step": 2235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788428521382809e-06,
+      "loss": 2.7023,
+      "step": 2236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788177922152682e-06,
+      "loss": 2.3896,
+      "step": 2237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787927177808955e-06,
+      "loss": 2.5492,
+      "step": 2238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787676288359225e-06,
+      "loss": 2.2801,
+      "step": 2239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787425253811097e-06,
+      "loss": 2.4861,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78717407417218e-06,
+      "loss": 2.3833,
+      "step": 2241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786922749450088e-06,
+      "loss": 2.2462,
+      "step": 2242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786671279652433e-06,
+      "loss": 2.5168,
+      "step": 2243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78641966478684e-06,
+      "loss": 2.3399,
+      "step": 2244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786167904860931e-06,
+      "loss": 2.2989,
+      "step": 2245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78591599988234e-06,
+      "loss": 2.5681,
+      "step": 2246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785663949858699e-06,
+      "loss": 2.462,
+      "step": 2247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785411754797647e-06,
+      "loss": 2.7861,
+      "step": 2248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785159414706828e-06,
+      "loss": 2.427,
+      "step": 2249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784906929593892e-06,
+      "loss": 2.5253,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784654299466485e-06,
+      "loss": 2.3747,
+      "step": 2251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784401524332268e-06,
+      "loss": 2.4173,
+      "step": 2252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7841486041989e-06,
+      "loss": 2.6828,
+      "step": 2253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783895539074046e-06,
+      "loss": 2.5024,
+      "step": 2254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783642328965376e-06,
+      "loss": 2.3724,
+      "step": 2255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783388973880564e-06,
+      "loss": 2.2719,
+      "step": 2256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78313547382729e-06,
+      "loss": 2.2422,
+      "step": 2257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782881828813233e-06,
+      "loss": 2.3751,
+      "step": 2258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782628038846083e-06,
+      "loss": 2.5065,
+      "step": 2259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782374103933531e-06,
+      "loss": 2.3624,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782120024083273e-06,
+      "loss": 2.509,
+      "step": 2261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781865799303008e-06,
+      "loss": 2.6538,
+      "step": 2262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781611429600442e-06,
+      "loss": 2.3303,
+      "step": 2263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781356914983284e-06,
+      "loss": 2.1889,
+      "step": 2264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781102255459247e-06,
+      "loss": 2.3773,
+      "step": 2265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780847451036049e-06,
+      "loss": 2.7697,
+      "step": 2266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780592501721414e-06,
+      "loss": 2.225,
+      "step": 2267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780337407523063e-06,
+      "loss": 2.48,
+      "step": 2268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780082168448735e-06,
+      "loss": 2.5593,
+      "step": 2269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779826784506161e-06,
+      "loss": 2.4958,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77957125570308e-06,
+      "loss": 2.3001,
+      "step": 2271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779315582047239e-06,
+      "loss": 2.5493,
+      "step": 2272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779059763546384e-06,
+      "loss": 2.4072,
+      "step": 2273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77880380020827e-06,
+      "loss": 2.7016,
+      "step": 2274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778547692040651e-06,
+      "loss": 2.4898,
+      "step": 2275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778291439051295e-06,
+      "loss": 2.3509,
+      "step": 2276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778035041247964e-06,
+      "loss": 2.3136,
+      "step": 2277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777778498638428e-06,
+      "loss": 2.3658,
+      "step": 2278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777521811230462e-06,
+      "loss": 2.4995,
+      "step": 2279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777264979031848e-06,
+      "loss": 2.7127,
+      "step": 2280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777008002050367e-06,
+      "loss": 2.5035,
+      "step": 2281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776750880293811e-06,
+      "loss": 2.5884,
+      "step": 2282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776493613769966e-06,
+      "loss": 2.3198,
+      "step": 2283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776236202486634e-06,
+      "loss": 2.4778,
+      "step": 2284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775978646451616e-06,
+      "loss": 2.5176,
+      "step": 2285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775720945672716e-06,
+      "loss": 2.7929,
+      "step": 2286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775463100157743e-06,
+      "loss": 2.4054,
+      "step": 2287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775205109914513e-06,
+      "loss": 2.598,
+      "step": 2288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774946974950846e-06,
+      "loss": 2.4594,
+      "step": 2289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774688695274564e-06,
+      "loss": 2.2618,
+      "step": 2290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774430270893494e-06,
+      "loss": 2.1333,
+      "step": 2291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774171701815468e-06,
+      "loss": 2.3291,
+      "step": 2292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773912988048326e-06,
+      "loss": 2.3264,
+      "step": 2293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773654129599904e-06,
+      "loss": 2.3642,
+      "step": 2294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773395126478047e-06,
+      "loss": 2.4459,
+      "step": 2295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77313597869061e-06,
+      "loss": 2.2099,
+      "step": 2296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77287668624544e-06,
+      "loss": 2.5178,
+      "step": 2297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772617249150398e-06,
+      "loss": 2.5327,
+      "step": 2298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772357667413348e-06,
+      "loss": 2.4282,
+      "step": 2299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772097941042159e-06,
+      "loss": 2.452,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771838070044698e-06,
+      "loss": 2.5212,
+      "step": 2301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771578054428841e-06,
+      "loss": 2.5572,
+      "step": 2302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77131789420247e-06,
+      "loss": 2.2662,
+      "step": 2303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77105758937347e-06,
+      "loss": 2.3262,
+      "step": 2304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77079713994973e-06,
+      "loss": 2.3062,
+      "step": 2305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77053654593914e-06,
+      "loss": 2.4386,
+      "step": 2306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770275807349603e-06,
+      "loss": 2.2295,
+      "step": 2307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770014924189017e-06,
+      "loss": 2.5623,
+      "step": 2308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769753896465291e-06,
+      "loss": 2.7538,
+      "step": 2309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769492724186334e-06,
+      "loss": 2.6392,
+      "step": 2310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769231407360063e-06,
+      "loss": 2.4228,
+      "step": 2311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768969945994396e-06,
+      "loss": 2.4854,
+      "step": 2312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768708340097258e-06,
+      "loss": 2.5812,
+      "step": 2313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768446589676576e-06,
+      "loss": 2.5371,
+      "step": 2314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768184694740284e-06,
+      "loss": 2.3845,
+      "step": 2315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76792265529632e-06,
+      "loss": 2.5249,
+      "step": 2316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767660471352624e-06,
+      "loss": 2.253,
+      "step": 2317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767398142917142e-06,
+      "loss": 2.566,
+      "step": 2318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767135669997824e-06,
+      "loss": 2.5856,
+      "step": 2319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766873052602627e-06,
+      "loss": 2.4745,
+      "step": 2320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76661029073951e-06,
+      "loss": 2.5405,
+      "step": 2321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766347384416432e-06,
+      "loss": 2.6085,
+      "step": 2322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766084333641366e-06,
+      "loss": 2.3963,
+      "step": 2323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76582113842228e-06,
+      "loss": 2.3158,
+      "step": 2324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765557798767154e-06,
+      "loss": 2.4972,
+      "step": 2325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765294314683966e-06,
+      "loss": 2.5165,
+      "step": 2326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765030686180704e-06,
+      "loss": 2.6277,
+      "step": 2327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764766913265356e-06,
+      "loss": 2.448,
+      "step": 2328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764502995945917e-06,
+      "loss": 2.3957,
+      "step": 2329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764238934230384e-06,
+      "loss": 2.5535,
+      "step": 2330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763974728126761e-06,
+      "loss": 2.3582,
+      "step": 2331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763710377643055e-06,
+      "loss": 2.6108,
+      "step": 2332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763445882787279e-06,
+      "loss": 2.3293,
+      "step": 2333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763181243567446e-06,
+      "loss": 2.5512,
+      "step": 2334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762916459991577e-06,
+      "loss": 2.2232,
+      "step": 2335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7626515320677e-06,
+      "loss": 2.6538,
+      "step": 2336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76238645980384e-06,
+      "loss": 2.4505,
+      "step": 2337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762121243208031e-06,
+      "loss": 2.343,
+      "step": 2338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761855882288314e-06,
+      "loss": 2.4787,
+      "step": 2339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761590377052728e-06,
+      "loss": 2.4652,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76132472750932e-06,
+      "loss": 2.5211,
+      "step": 2341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761058933666143e-06,
+      "loss": 2.5003,
+      "step": 2342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76079299553125e-06,
+      "loss": 2.4405,
+      "step": 2343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760526913112702e-06,
+      "loss": 2.226,
+      "step": 2344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760260686418564e-06,
+      "loss": 2.3858,
+      "step": 2345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759994315456901e-06,
+      "loss": 2.4676,
+      "step": 2346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75972780023579e-06,
+      "loss": 2.4429,
+      "step": 2347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759461140763303e-06,
+      "loss": 2.5304,
+      "step": 2348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759194337047527e-06,
+      "loss": 2.2685,
+      "step": 2349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758927389096546e-06,
+      "loss": 2.3685,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758660296918447e-06,
+      "loss": 2.6558,
+      "step": 2351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758393060521331e-06,
+      "loss": 2.5949,
+      "step": 2352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758125679913292e-06,
+      "loss": 2.4099,
+      "step": 2353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757858155102436e-06,
+      "loss": 2.4312,
+      "step": 2354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757590486096868e-06,
+      "loss": 2.4133,
+      "step": 2355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757322672904702e-06,
+      "loss": 2.4567,
+      "step": 2356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757054715534056e-06,
+      "loss": 2.5117,
+      "step": 2357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756786613993048e-06,
+      "loss": 2.9169,
+      "step": 2358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756518368289805e-06,
+      "loss": 2.4098,
+      "step": 2359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756249978432457e-06,
+      "loss": 2.2616,
+      "step": 2360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755981444429137e-06,
+      "loss": 2.5488,
+      "step": 2361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755712766287983e-06,
+      "loss": 2.3786,
+      "step": 2362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755443944017138e-06,
+      "loss": 2.4154,
+      "step": 2363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75517497762475e-06,
+      "loss": 2.5179,
+      "step": 2364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754905867118969e-06,
+      "loss": 2.5695,
+      "step": 2365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754636612507952e-06,
+      "loss": 2.3298,
+      "step": 2366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75436721379986e-06,
+      "loss": 2.3378,
+      "step": 2367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754097671002855e-06,
+      "loss": 2.3667,
+      "step": 2368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75382798412511e-06,
+      "loss": 2.1729,
+      "step": 2369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753558153174793e-06,
+      "loss": 2.5173,
+      "step": 2370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753288178160086e-06,
+      "loss": 2.6554,
+      "step": 2371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753018059089168e-06,
+      "loss": 2.2894,
+      "step": 2372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75274779597023e-06,
+      "loss": 2.3864,
+      "step": 2373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752477388811456e-06,
+      "loss": 2.4863,
+      "step": 2374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752206837621045e-06,
+      "loss": 2.2831,
+      "step": 2375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751936142407198e-06,
+      "loss": 2.3815,
+      "step": 2376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751665303178118e-06,
+      "loss": 2.5249,
+      "step": 2377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75139431994201e-06,
+      "loss": 2.5181,
+      "step": 2378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751123192707089e-06,
+      "loss": 2.4554,
+      "step": 2379
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750851921481573e-06,
+      "loss": 2.29,
+      "step": 2380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750580506273681e-06,
+      "loss": 2.4879,
+      "step": 2381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75030894709164e-06,
+      "loss": 2.4454,
+      "step": 2382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750037243943682e-06,
+      "loss": 2.4663,
+      "step": 2383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749765396838038e-06,
+      "loss": 2.2698,
+      "step": 2384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749493405782947e-06,
+      "loss": 2.4235,
+      "step": 2385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749221270786656e-06,
+      "loss": 2.4713,
+      "step": 2386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748948991857408e-06,
+      "loss": 2.5936,
+      "step": 2387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748676569003455e-06,
+      "loss": 2.2492,
+      "step": 2388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748404002233057e-06,
+      "loss": 2.541,
+      "step": 2389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748131291554472e-06,
+      "loss": 2.4241,
+      "step": 2390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747858436975965e-06,
+      "loss": 2.3557,
+      "step": 2391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747585438505806e-06,
+      "loss": 2.4957,
+      "step": 2392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747312296152268e-06,
+      "loss": 2.3176,
+      "step": 2393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74703900992363e-06,
+      "loss": 2.4978,
+      "step": 2394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746765579828173e-06,
+      "loss": 2.3835,
+      "step": 2395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746492005874186e-06,
+      "loss": 2.4668,
+      "step": 2396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746218288069957e-06,
+      "loss": 2.8098,
+      "step": 2397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745944426423783e-06,
+      "loss": 2.5756,
+      "step": 2398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745670420943965e-06,
+      "loss": 2.4973,
+      "step": 2399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745396271638805e-06,
+      "loss": 2.5586,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745121978516613e-06,
+      "loss": 2.4839,
+      "step": 2401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744847541585703e-06,
+      "loss": 2.5057,
+      "step": 2402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744572960854389e-06,
+      "loss": 2.2671,
+      "step": 2403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744298236330993e-06,
+      "loss": 2.4873,
+      "step": 2404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744023368023845e-06,
+      "loss": 2.3174,
+      "step": 2405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743748355941272e-06,
+      "loss": 2.5884,
+      "step": 2406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74347320009161e-06,
+      "loss": 2.5234,
+      "step": 2407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743197900483196e-06,
+      "loss": 2.3765,
+      "step": 2408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742922457124377e-06,
+      "loss": 2.5874,
+      "step": 2409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742646870023498e-06,
+      "loss": 2.4578,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74237113918891e-06,
+      "loss": 2.3563,
+      "step": 2411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742095264628972e-06,
+      "loss": 2.2702,
+      "step": 2412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741819246352046e-06,
+      "loss": 2.4196,
+      "step": 2413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741543084366493e-06,
+      "loss": 2.3273,
+      "step": 2414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741266778680686e-06,
+      "loss": 2.4076,
+      "step": 2415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740990329302998e-06,
+      "loss": 2.4978,
+      "step": 2416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740713736241805e-06,
+      "loss": 2.2877,
+      "step": 2417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740436999505494e-06,
+      "loss": 2.4509,
+      "step": 2418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740160119102447e-06,
+      "loss": 2.4748,
+      "step": 2419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73988309504106e-06,
+      "loss": 2.5659,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739605927329725e-06,
+      "loss": 2.342,
+      "step": 2421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739328615976845e-06,
+      "loss": 2.6282,
+      "step": 2422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739051160990823e-06,
+      "loss": 2.4187,
+      "step": 2423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738773562380066e-06,
+      "loss": 2.3124,
+      "step": 2424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738495820152988e-06,
+      "loss": 2.3979,
+      "step": 2425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738217934318011e-06,
+      "loss": 2.2771,
+      "step": 2426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73793990488355e-06,
+      "loss": 2.6455,
+      "step": 2427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737661731858035e-06,
+      "loss": 2.1482,
+      "step": 2428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737383415249894e-06,
+      "loss": 2.6681,
+      "step": 2429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737104955067566e-06,
+      "loss": 2.3376,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736826351319486e-06,
+      "loss": 2.3413,
+      "step": 2431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7365476040141e-06,
+      "loss": 2.7365,
+      "step": 2432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736268713159854e-06,
+      "loss": 2.2116,
+      "step": 2433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735989678765202e-06,
+      "loss": 2.4307,
+      "step": 2434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7357105008386e-06,
+      "loss": 2.6938,
+      "step": 2435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735431179388508e-06,
+      "loss": 2.3206,
+      "step": 2436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735151714423392e-06,
+      "loss": 2.4735,
+      "step": 2437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734872105951723e-06,
+      "loss": 2.4373,
+      "step": 2438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734592353981973e-06,
+      "loss": 2.5163,
+      "step": 2439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734312458522621e-06,
+      "loss": 2.4299,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73403241958215e-06,
+      "loss": 2.3461,
+      "step": 2441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733752237169048e-06,
+      "loss": 2.3285,
+      "step": 2442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733471911291802e-06,
+      "loss": 2.3109,
+      "step": 2443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733191441958914e-06,
+      "loss": 2.2649,
+      "step": 2444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73291082917888e-06,
+      "loss": 2.3376,
+      "step": 2445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732630072960207e-06,
+      "loss": 2.5535,
+      "step": 2446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7323491733114e-06,
+      "loss": 2.1323,
+      "step": 2447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732068130240976e-06,
+      "loss": 2.3574,
+      "step": 2448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731786943757451e-06,
+      "loss": 2.4337,
+      "step": 2449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731505613869346e-06,
+      "loss": 2.5481,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731224140585187e-06,
+      "loss": 2.3282,
+      "step": 2451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730942523913505e-06,
+      "loss": 2.2624,
+      "step": 2452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730660763862837e-06,
+      "loss": 2.4937,
+      "step": 2453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73037886044172e-06,
+      "loss": 2.3369,
+      "step": 2454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730096813658698e-06,
+      "loss": 2.5457,
+      "step": 2455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729814623522318e-06,
+      "loss": 2.3952,
+      "step": 2456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729532290041133e-06,
+      "loss": 2.5159,
+      "step": 2457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7292498132237e-06,
+      "loss": 2.6073,
+      "step": 2458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728967193078581e-06,
+      "loss": 2.364,
+      "step": 2459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728684429614338e-06,
+      "loss": 2.4186,
+      "step": 2460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728401522839543e-06,
+      "loss": 2.2867,
+      "step": 2461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728118472762771e-06,
+      "loss": 2.551,
+      "step": 2462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727835279392597e-06,
+      "loss": 2.4128,
+      "step": 2463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727551942737605e-06,
+      "loss": 2.4823,
+      "step": 2464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727268462806383e-06,
+      "loss": 2.4266,
+      "step": 2465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726984839607522e-06,
+      "loss": 2.4331,
+      "step": 2466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726701073149619e-06,
+      "loss": 2.4385,
+      "step": 2467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726417163441269e-06,
+      "loss": 2.6548,
+      "step": 2468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726133110491081e-06,
+      "loss": 2.2396,
+      "step": 2469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725848914307662e-06,
+      "loss": 2.5128,
+      "step": 2470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725564574899627e-06,
+      "loss": 2.4472,
+      "step": 2471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72528009227559e-06,
+      "loss": 2.4607,
+      "step": 2472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724995466444175e-06,
+      "loss": 2.6087,
+      "step": 2473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724710697414009e-06,
+      "loss": 2.5422,
+      "step": 2474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72442578519372e-06,
+      "loss": 2.5251,
+      "step": 2475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724140729791942e-06,
+      "loss": 2.5066,
+      "step": 2476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723855531217316e-06,
+      "loss": 2.5333,
+      "step": 2477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723570189478488e-06,
+      "loss": 2.2512,
+      "step": 2478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723284704584101e-06,
+      "loss": 2.4586,
+      "step": 2479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722999076542809e-06,
+      "loss": 2.2648,
+      "step": 2480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72271330536327e-06,
+      "loss": 2.5787,
+      "step": 2481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722427391054142e-06,
+      "loss": 2.4368,
+      "step": 2482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72214133362409e-06,
+      "loss": 2.4932,
+      "step": 2483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721855133081786e-06,
+      "loss": 2.1664,
+      "step": 2484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721568789435902e-06,
+      "loss": 2.5057,
+      "step": 2485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721282302695117e-06,
+      "loss": 2.5634,
+      "step": 2486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720995672868115e-06,
+      "loss": 2.3883,
+      "step": 2487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720708899963577e-06,
+      "loss": 2.5327,
+      "step": 2488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720421983990201e-06,
+      "loss": 2.3993,
+      "step": 2489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720134924956678e-06,
+      "loss": 2.3637,
+      "step": 2490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71984772287171e-06,
+      "loss": 2.3806,
+      "step": 2491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719560377744e-06,
+      "loss": 2.1333,
+      "step": 2492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719272889582257e-06,
+      "loss": 2.4947,
+      "step": 2493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718985258395195e-06,
+      "loss": 2.46,
+      "step": 2494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718697484191528e-06,
+      "loss": 2.4234,
+      "step": 2495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71840956697998e-06,
+      "loss": 2.5364,
+      "step": 2496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718121506769275e-06,
+      "loss": 2.1715,
+      "step": 2497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717833303568145e-06,
+      "loss": 2.372,
+      "step": 2498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717544957385325e-06,
+      "loss": 2.6796,
+      "step": 2499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71725646822955e-06,
+      "loss": 2.4614,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716967836109568e-06,
+      "loss": 2.0626,
+      "step": 2501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716679061034122e-06,
+      "loss": 2.518,
+      "step": 2502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716390143011967e-06,
+      "loss": 2.6445,
+      "step": 2503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716101082051857e-06,
+      "loss": 2.6706,
+      "step": 2504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715811878162556e-06,
+      "loss": 2.3861,
+      "step": 2505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715522531352823e-06,
+      "loss": 2.5359,
+      "step": 2506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715233041631434e-06,
+      "loss": 2.2954,
+      "step": 2507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714943409007157e-06,
+      "loss": 2.6242,
+      "step": 2508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71465363348877e-06,
+      "loss": 2.5698,
+      "step": 2509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714363715085059e-06,
+      "loss": 2.5379,
+      "step": 2510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714073653804808e-06,
+      "loss": 2.3066,
+      "step": 2511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713783449656807e-06,
+      "loss": 2.7404,
+      "step": 2512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713493102649853e-06,
+      "loss": 2.2728,
+      "step": 2513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713202612792744e-06,
+      "loss": 2.2549,
+      "step": 2514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712911980094285e-06,
+      "loss": 2.4941,
+      "step": 2515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712621204563285e-06,
+      "loss": 2.3678,
+      "step": 2516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71233028620855e-06,
+      "loss": 2.5851,
+      "step": 2517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712039225038907e-06,
+      "loss": 2.4105,
+      "step": 2518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71174802106317e-06,
+      "loss": 2.5523,
+      "step": 2519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711456674290166e-06,
+      "loss": 2.5699,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711165184728723e-06,
+      "loss": 2.3868,
+      "step": 2521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71087355238768e-06,
+      "loss": 2.455,
+      "step": 2522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71058177727587e-06,
+      "loss": 2.3141,
+      "step": 2523
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71028985940214e-06,
+      "loss": 2.6027,
+      "step": 2524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709997798775334e-06,
+      "loss": 2.4637,
+      "step": 2525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709705595404304e-06,
+      "loss": 2.3305,
+      "step": 2526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709413249297908e-06,
+      "loss": 2.5033,
+      "step": 2527
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709120760465002e-06,
+      "loss": 2.2354,
+      "step": 2528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708828128914455e-06,
+      "loss": 2.4581,
+      "step": 2529
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708535354655132e-06,
+      "loss": 2.3873,
+      "step": 2530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708242437695908e-06,
+      "loss": 2.2498,
+      "step": 2531
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70794937804566e-06,
+      "loss": 2.6506,
+      "step": 2532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707656175713266e-06,
+      "loss": 2.6881,
+      "step": 2533
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707362830707618e-06,
+      "loss": 2.5436,
+      "step": 2534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707069343037604e-06,
+      "loss": 2.4214,
+      "step": 2535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706775712712117e-06,
+      "loss": 2.1776,
+      "step": 2536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706481939740059e-06,
+      "loss": 2.1957,
+      "step": 2537
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70618802413033e-06,
+      "loss": 2.4718,
+      "step": 2538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705893965891838e-06,
+      "loss": 2.565,
+      "step": 2539
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705599765033498e-06,
+      "loss": 2.3247,
+      "step": 2540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705305421564222e-06,
+      "loss": 2.4839,
+      "step": 2541
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705010935492933e-06,
+      "loss": 2.3143,
+      "step": 2542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704716306828557e-06,
+      "loss": 2.2524,
+      "step": 2543
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704421535580022e-06,
+      "loss": 2.4891,
+      "step": 2544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70412662175626e-06,
+      "loss": 2.6762,
+      "step": 2545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703831565366212e-06,
+      "loss": 2.5437,
+      "step": 2546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703536366418816e-06,
+      "loss": 2.3285,
+      "step": 2547
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703241024923022e-06,
+      "loss": 2.4049,
+      "step": 2548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70294554088778e-06,
+      "loss": 2.3159,
+      "step": 2549
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702649914322044e-06,
+      "loss": 2.5519,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702354145234774e-06,
+      "loss": 2.7903,
+      "step": 2551
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702058233634936e-06,
+      "loss": 2.0334,
+      "step": 2552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701762179531495e-06,
+      "loss": 2.2301,
+      "step": 2553
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701465982933423e-06,
+      "loss": 2.2992,
+      "step": 2554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701169643849699e-06,
+      "loss": 2.4757,
+      "step": 2555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700873162289305e-06,
+      "loss": 2.4391,
+      "step": 2556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700576538261224e-06,
+      "loss": 2.6078,
+      "step": 2557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700279771774446e-06,
+      "loss": 2.4644,
+      "step": 2558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699982862837965e-06,
+      "loss": 2.376,
+      "step": 2559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699685811460779e-06,
+      "loss": 2.5336,
+      "step": 2560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699388617651891e-06,
+      "loss": 2.5923,
+      "step": 2561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69909128142031e-06,
+      "loss": 2.3008,
+      "step": 2562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698793802775044e-06,
+      "loss": 2.4578,
+      "step": 2563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69849618172511e-06,
+      "loss": 2.389,
+      "step": 2564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698198418279529e-06,
+      "loss": 2.5833,
+      "step": 2565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697900512447324e-06,
+      "loss": 2.314,
+      "step": 2566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697602464237524e-06,
+      "loss": 2.3718,
+      "step": 2567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697304273659161e-06,
+      "loss": 2.4043,
+      "step": 2568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697005940721272e-06,
+      "loss": 2.35,
+      "step": 2569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6967074654329e-06,
+      "loss": 2.3763,
+      "step": 2570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69640884780309e-06,
+      "loss": 2.2026,
+      "step": 2571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69611008784089e-06,
+      "loss": 2.2534,
+      "step": 2572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695811185555359e-06,
+      "loss": 2.6125,
+      "step": 2573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695512140955552e-06,
+      "loss": 2.5051,
+      "step": 2574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695212954050532e-06,
+      "loss": 2.5692,
+      "step": 2575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69491362484937e-06,
+      "loss": 2.3065,
+      "step": 2576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694614153361134e-06,
+      "loss": 2.3766,
+      "step": 2577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694314539594902e-06,
+      "loss": 2.3816,
+      "step": 2578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694014783559751e-06,
+      "loss": 2.3862,
+      "step": 2579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69371488526477e-06,
+      "loss": 2.3451,
+      "step": 2580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693414844719047e-06,
+      "loss": 2.5366,
+      "step": 2581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693114661931673e-06,
+      "loss": 2.4694,
+      "step": 2582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692814336911747e-06,
+      "loss": 2.4313,
+      "step": 2583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69251386966837e-06,
+      "loss": 2.4461,
+      "step": 2584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692213260210648e-06,
+      "loss": 2.2928,
+      "step": 2585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691912508547694e-06,
+      "loss": 2.5186,
+      "step": 2586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69161161468862e-06,
+      "loss": 2.6097,
+      "step": 2587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691310578642546e-06,
+      "loss": 2.4178,
+      "step": 2588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691009400418595e-06,
+      "loss": 2.2728,
+      "step": 2589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690708080025896e-06,
+      "loss": 2.5221,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69040661747358e-06,
+      "loss": 2.5345,
+      "step": 2591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690105012770783e-06,
+      "loss": 2.3528,
+      "step": 2592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689803265926648e-06,
+      "loss": 2.4556,
+      "step": 2593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689501376950316e-06,
+      "loss": 2.4661,
+      "step": 2594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68919934585094e-06,
+      "loss": 2.3333,
+      "step": 2595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68889717263767e-06,
+      "loss": 2.3824,
+      "step": 2596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688594857319665e-06,
+      "loss": 2.5719,
+      "step": 2597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688292399906089e-06,
+      "loss": 2.4752,
+      "step": 2598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687989800406109e-06,
+      "loss": 2.476,
+      "step": 2599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687687058828891e-06,
+      "loss": 2.4915,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687384175183616e-06,
+      "loss": 2.4426,
+      "step": 2601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687081149479459e-06,
+      "loss": 2.3195,
+      "step": 2602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686777981725606e-06,
+      "loss": 2.4234,
+      "step": 2603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686474671931244e-06,
+      "loss": 2.6386,
+      "step": 2604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686171220105565e-06,
+      "loss": 2.4378,
+      "step": 2605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685867626257765e-06,
+      "loss": 2.6241,
+      "step": 2606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685563890397048e-06,
+      "loss": 2.3802,
+      "step": 2607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685260012532617e-06,
+      "loss": 2.3654,
+      "step": 2608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684955992673681e-06,
+      "loss": 2.3461,
+      "step": 2609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684651830829456e-06,
+      "loss": 2.5407,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684347527009156e-06,
+      "loss": 2.3966,
+      "step": 2611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68404308122201e-06,
+      "loss": 2.403,
+      "step": 2612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683738493477239e-06,
+      "loss": 2.1688,
+      "step": 2613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683433763784076e-06,
+      "loss": 2.2926,
+      "step": 2614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683128892151757e-06,
+      "loss": 2.5413,
+      "step": 2615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682823878589518e-06,
+      "loss": 2.5965,
+      "step": 2616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682518723106608e-06,
+      "loss": 2.9309,
+      "step": 2617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682213425712274e-06,
+      "loss": 2.4581,
+      "step": 2618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681907986415767e-06,
+      "loss": 2.3661,
+      "step": 2619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681602405226343e-06,
+      "loss": 2.5272,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681296682153265e-06,
+      "loss": 2.5239,
+      "step": 2621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6809908172058e-06,
+      "loss": 2.557,
+      "step": 2622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680684810393214e-06,
+      "loss": 2.463,
+      "step": 2623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680378661724782e-06,
+      "loss": 2.37,
+      "step": 2624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680072371209786e-06,
+      "loss": 2.3733,
+      "step": 2625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679765938857503e-06,
+      "loss": 2.4913,
+      "step": 2626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679459364677224e-06,
+      "loss": 2.4443,
+      "step": 2627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67915264867824e-06,
+      "loss": 2.3494,
+      "step": 2628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678845790869844e-06,
+      "loss": 2.5069,
+      "step": 2629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678538791261336e-06,
+      "loss": 2.4911,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678231649862024e-06,
+      "loss": 2.4478,
+      "step": 2631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677924366681213e-06,
+      "loss": 2.4051,
+      "step": 2632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677616941728216e-06,
+      "loss": 2.3653,
+      "step": 2633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677309375012351e-06,
+      "loss": 2.3906,
+      "step": 2634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67700166654294e-06,
+      "loss": 2.474,
+      "step": 2635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676693816329307e-06,
+      "loss": 2.6371,
+      "step": 2636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676385824380782e-06,
+      "loss": 2.4733,
+      "step": 2637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6760776907067e-06,
+      "loss": 2.6165,
+      "step": 2638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675769415316398e-06,
+      "loss": 2.6006,
+      "step": 2639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675460998219222e-06,
+      "loss": 2.3922,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675152439424515e-06,
+      "loss": 2.3384,
+      "step": 2641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67484373894163e-06,
+      "loss": 2.1873,
+      "step": 2642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674534896779926e-06,
+      "loss": 2.3969,
+      "step": 2643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674225912948758e-06,
+      "loss": 2.4476,
+      "step": 2644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673916787457492e-06,
+      "loss": 2.2703,
+      "step": 2645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673607520315498e-06,
+      "loss": 2.2461,
+      "step": 2646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673298111532147e-06,
+      "loss": 2.4965,
+      "step": 2647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672988561116818e-06,
+      "loss": 2.5084,
+      "step": 2648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67267886907889e-06,
+      "loss": 2.4084,
+      "step": 2649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67236903542775e-06,
+      "loss": 2.4322,
+      "step": 2650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67205906017279e-06,
+      "loss": 2.3874,
+      "step": 2651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6717489433234e-06,
+      "loss": 2.5392,
+      "step": 2652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67143868488898e-06,
+      "loss": 2.3687,
+      "step": 2653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671128284878937e-06,
+      "loss": 2.5209,
+      "step": 2654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670817743302674e-06,
+      "loss": 2.4581,
+      "step": 2655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670507060169603e-06,
+      "loss": 2.5467,
+      "step": 2656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67019623548914e-06,
+      "loss": 2.4242,
+      "step": 2657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669885269270706e-06,
+      "loss": 2.4709,
+      "step": 2658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669574161523723e-06,
+      "loss": 2.2893,
+      "step": 2659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669262912257623e-06,
+      "loss": 2.3321,
+      "step": 2660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668951521481837e-06,
+      "loss": 2.4207,
+      "step": 2661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668639989205802e-06,
+      "loss": 2.3484,
+      "step": 2662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668328315438961e-06,
+      "loss": 2.5338,
+      "step": 2663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668016500190758e-06,
+      "loss": 2.4763,
+      "step": 2664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667704543470644e-06,
+      "loss": 2.3072,
+      "step": 2665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667392445288071e-06,
+      "loss": 2.2732,
+      "step": 2666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667080205652501e-06,
+      "loss": 2.3511,
+      "step": 2667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666767824573398e-06,
+      "loss": 2.37,
+      "step": 2668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666455302060224e-06,
+      "loss": 2.5472,
+      "step": 2669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666142638122456e-06,
+      "loss": 2.5315,
+      "step": 2670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665829832769564e-06,
+      "loss": 1.946,
+      "step": 2671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665516886011035e-06,
+      "loss": 2.4392,
+      "step": 2672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665203797856346e-06,
+      "loss": 2.3567,
+      "step": 2673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664890568314993e-06,
+      "loss": 2.0906,
+      "step": 2674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664577197396464e-06,
+      "loss": 2.2185,
+      "step": 2675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664263685110255e-06,
+      "loss": 2.6241,
+      "step": 2676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663950031465872e-06,
+      "loss": 2.5307,
+      "step": 2677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66363623647282e-06,
+      "loss": 2.1931,
+      "step": 2678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663322300140605e-06,
+      "loss": 2.3474,
+      "step": 2679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663008222478747e-06,
+      "loss": 2.482,
+      "step": 2680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66269400349676e-06,
+      "loss": 2.3142,
+      "step": 2681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662379643204169e-06,
+      "loss": 2.3197,
+      "step": 2682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662065141610501e-06,
+      "loss": 2.2917,
+      "step": 2683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661750498725289e-06,
+      "loss": 2.417,
+      "step": 2684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661435714558066e-06,
+      "loss": 2.3635,
+      "step": 2685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661120789118373e-06,
+      "loss": 2.531,
+      "step": 2686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660805722415755e-06,
+      "loss": 2.3399,
+      "step": 2687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660490514459761e-06,
+      "loss": 2.5909,
+      "step": 2688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660175165259942e-06,
+      "loss": 2.3106,
+      "step": 2689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659859674825858e-06,
+      "loss": 2.4966,
+      "step": 2690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659544043167066e-06,
+      "loss": 2.4926,
+      "step": 2691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659228270293137e-06,
+      "loss": 2.3941,
+      "step": 2692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658912356213637e-06,
+      "loss": 2.2308,
+      "step": 2693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658596300938143e-06,
+      "loss": 2.282,
+      "step": 2694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658280104476231e-06,
+      "loss": 2.2487,
+      "step": 2695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657963766837486e-06,
+      "loss": 2.6085,
+      "step": 2696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657647288031494e-06,
+      "loss": 2.6514,
+      "step": 2697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657330668067849e-06,
+      "loss": 2.128,
+      "step": 2698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65701390695614e-06,
+      "loss": 2.3805,
+      "step": 2699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656697004705975e-06,
+      "loss": 2.5467,
+      "step": 2700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656379961326954e-06,
+      "loss": 2.4837,
+      "step": 2701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656062776828685e-06,
+      "loss": 2.4452,
+      "step": 2702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655745451220785e-06,
+      "loss": 2.7257,
+      "step": 2703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655427984512864e-06,
+      "loss": 2.5362,
+      "step": 2704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65511037671455e-06,
+      "loss": 2.4755,
+      "step": 2705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654792627835465e-06,
+      "loss": 2.3578,
+      "step": 2706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654474737885242e-06,
+      "loss": 2.5507,
+      "step": 2707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65415670687351e-06,
+      "loss": 2.4931,
+      "step": 2708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653838534809911e-06,
+      "loss": 2.4064,
+      "step": 2709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65352022170409e-06,
+      "loss": 2.4696,
+      "step": 2710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65320176756569e-06,
+      "loss": 2.1948,
+      "step": 2711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652883172404363e-06,
+      "loss": 2.3126,
+      "step": 2712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652564436229765e-06,
+      "loss": 2.6242,
+      "step": 2713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652245559051556e-06,
+      "loss": 2.3447,
+      "step": 2714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651926540879401e-06,
+      "loss": 2.2591,
+      "step": 2715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651607381722967e-06,
+      "loss": 2.0572,
+      "step": 2716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651288081591927e-06,
+      "loss": 2.1489,
+      "step": 2717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650968640495959e-06,
+      "loss": 2.2826,
+      "step": 2718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650649058444741e-06,
+      "loss": 2.455,
+      "step": 2719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650329335447962e-06,
+      "loss": 2.3512,
+      "step": 2720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65000947151531e-06,
+      "loss": 2.4003,
+      "step": 2721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64968946665648e-06,
+      "loss": 2.5599,
+      "step": 2722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64936932088117e-06,
+      "loss": 2.8061,
+      "step": 2723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64904903419908e-06,
+      "loss": 2.3628,
+      "step": 2724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64872860661992e-06,
+      "loss": 2.4186,
+      "step": 2725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6484080381534e-06,
+      "loss": 2.5554,
+      "step": 2726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648087328809236e-06,
+      "loss": 2.4915,
+      "step": 2727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647766478597147e-06,
+      "loss": 2.3715,
+      "step": 2728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647445487526857e-06,
+      "loss": 2.2692,
+      "step": 2729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647124355608096e-06,
+      "loss": 2.4104,
+      "step": 2730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64680308285059e-06,
+      "loss": 2.3466,
+      "step": 2731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646481669264084e-06,
+      "loss": 2.4714,
+      "step": 2732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646160114858315e-06,
+      "loss": 2.3918,
+      "step": 2733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645838419643027e-06,
+      "loss": 2.4643,
+      "step": 2734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64551658362797e-06,
+      "loss": 2.6355,
+      "step": 2735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645194606822902e-06,
+      "loss": 2.203,
+      "step": 2736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644872489237575e-06,
+      "loss": 2.6451,
+      "step": 2737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644550230881755e-06,
+      "loss": 2.4953,
+      "step": 2738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644227831765208e-06,
+      "loss": 2.4934,
+      "step": 2739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643905291897702e-06,
+      "loss": 2.4306,
+      "step": 2740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643582611289017e-06,
+      "loss": 2.464,
+      "step": 2741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64325978994893e-06,
+      "loss": 2.5844,
+      "step": 2742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642936827887223e-06,
+      "loss": 2.2289,
+      "step": 2743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642613725113688e-06,
+      "loss": 2.3782,
+      "step": 2744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642290481638113e-06,
+      "loss": 2.4021,
+      "step": 2745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641967097470295e-06,
+      "loss": 2.236,
+      "step": 2746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641643572620038e-06,
+      "loss": 2.4253,
+      "step": 2747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641319907097143e-06,
+      "loss": 2.5784,
+      "step": 2748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640996100911421e-06,
+      "loss": 2.3471,
+      "step": 2749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640672154072685e-06,
+      "loss": 2.2734,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640348066590755e-06,
+      "loss": 2.31,
+      "step": 2751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64002383847545e-06,
+      "loss": 2.5246,
+      "step": 2752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639699469736598e-06,
+      "loss": 2.3325,
+      "step": 2753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639374960384026e-06,
+      "loss": 2.2132,
+      "step": 2754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639050310427576e-06,
+      "loss": 2.5552,
+      "step": 2755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638725519877081e-06,
+      "loss": 2.1773,
+      "step": 2756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638400588742385e-06,
+      "loss": 2.6036,
+      "step": 2757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638075517033338e-06,
+      "loss": 2.3345,
+      "step": 2758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637750304759789e-06,
+      "loss": 2.6106,
+      "step": 2759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637424951931597e-06,
+      "loss": 2.3567,
+      "step": 2760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637099458558622e-06,
+      "loss": 2.5024,
+      "step": 2761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636773824650725e-06,
+      "loss": 2.3977,
+      "step": 2762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636448050217777e-06,
+      "loss": 2.5234,
+      "step": 2763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636122135269655e-06,
+      "loss": 2.3723,
+      "step": 2764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635796079816229e-06,
+      "loss": 2.5802,
+      "step": 2765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635469883867388e-06,
+      "loss": 2.4143,
+      "step": 2766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63514354743301e-06,
+      "loss": 2.2974,
+      "step": 2767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634817070522994e-06,
+      "loss": 2.5709,
+      "step": 2768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634490453147228e-06,
+      "loss": 2.5357,
+      "step": 2769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634163695315613e-06,
+      "loss": 2.5518,
+      "step": 2770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633836797038052e-06,
+      "loss": 2.2443,
+      "step": 2771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633509758324452e-06,
+      "loss": 2.3375,
+      "step": 2772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633182579184723e-06,
+      "loss": 2.44,
+      "step": 2773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632855259628784e-06,
+      "loss": 2.208,
+      "step": 2774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632527799666552e-06,
+      "loss": 2.5369,
+      "step": 2775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632200199307953e-06,
+      "loss": 2.3499,
+      "step": 2776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631872458562913e-06,
+      "loss": 2.6917,
+      "step": 2777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631544577441367e-06,
+      "loss": 2.6376,
+      "step": 2778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63121655595325e-06,
+      "loss": 2.5468,
+      "step": 2779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630888394108506e-06,
+      "loss": 2.3145,
+      "step": 2780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63056009191708e-06,
+      "loss": 2.3485,
+      "step": 2781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63023164938892e-06,
+      "loss": 2.2502,
+      "step": 2782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62990306653398e-06,
+      "loss": 2.0749,
+      "step": 2783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62957434336222e-06,
+      "loss": 2.4638,
+      "step": 2784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6292454798836e-06,
+      "loss": 2.6313,
+      "step": 2785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62891647610809e-06,
+      "loss": 2.7496,
+      "step": 2786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628587332045657e-06,
+      "loss": 2.564,
+      "step": 2787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628258047706282e-06,
+      "loss": 2.3953,
+      "step": 2788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627928623099938e-06,
+      "loss": 2.5216,
+      "step": 2789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627599058236613e-06,
+      "loss": 2.3208,
+      "step": 2790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627269353126295e-06,
+      "loss": 2.4123,
+      "step": 2791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626939507778972e-06,
+      "loss": 2.4212,
+      "step": 2792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626609522204645e-06,
+      "loss": 2.5804,
+      "step": 2793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626279396413314e-06,
+      "loss": 2.4163,
+      "step": 2794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625949130414982e-06,
+      "loss": 2.4507,
+      "step": 2795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62561872421966e-06,
+      "loss": 2.3571,
+      "step": 2796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625288177837362e-06,
+      "loss": 2.5833,
+      "step": 2797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624957491278104e-06,
+      "loss": 2.3972,
+      "step": 2798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624626664551908e-06,
+      "loss": 2.6177,
+      "step": 2799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624295697668802e-06,
+      "loss": 2.4121,
+      "step": 2800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623964590638816e-06,
+      "loss": 2.3153,
+      "step": 2801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623633343471983e-06,
+      "loss": 2.3371,
+      "step": 2802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623301956178344e-06,
+      "loss": 2.5583,
+      "step": 2803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622970428767941e-06,
+      "loss": 2.6161,
+      "step": 2804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622638761250825e-06,
+      "loss": 2.5214,
+      "step": 2805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622306953637042e-06,
+      "loss": 2.691,
+      "step": 2806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621975005936651e-06,
+      "loss": 2.3733,
+      "step": 2807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621642918159711e-06,
+      "loss": 2.4662,
+      "step": 2808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621310690316288e-06,
+      "loss": 2.4783,
+      "step": 2809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62097832241645e-06,
+      "loss": 2.3816,
+      "step": 2810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62064581447027e-06,
+      "loss": 2.4343,
+      "step": 2811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620313166487826e-06,
+      "loss": 2.0614,
+      "step": 2812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6199803784792e-06,
+      "loss": 2.443,
+      "step": 2813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619647450454473e-06,
+      "loss": 2.3364,
+      "step": 2814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61931438242374e-06,
+      "loss": 2.3358,
+      "step": 2815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618981174397093e-06,
+      "loss": 2.4759,
+      "step": 2816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618647826384632e-06,
+      "loss": 2.3497,
+      "step": 2817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618314338396458e-06,
+      "loss": 2.4519,
+      "step": 2818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617980710442678e-06,
+      "loss": 2.5343,
+      "step": 2819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617646942533403e-06,
+      "loss": 2.4109,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617313034678751e-06,
+      "loss": 2.244,
+      "step": 2821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616978986888838e-06,
+      "loss": 2.2943,
+      "step": 2822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61664479917379e-06,
+      "loss": 2.6136,
+      "step": 2823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616310471543735e-06,
+      "loss": 2.3028,
+      "step": 2824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615976004008804e-06,
+      "loss": 2.3348,
+      "step": 2825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615641396579136e-06,
+      "loss": 2.4701,
+      "step": 2826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61530664926487e-06,
+      "loss": 2.1511,
+      "step": 2827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61497176207615e-06,
+      "loss": 2.2736,
+      "step": 2828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61463673502313e-06,
+      "loss": 2.3501,
+      "step": 2829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614301568115956e-06,
+      "loss": 2.4645,
+      "step": 2830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613966261364791e-06,
+      "loss": 2.472,
+      "step": 2831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613630814779797e-06,
+      "loss": 2.4184,
+      "step": 2832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61329522837114e-06,
+      "loss": 2.4636,
+      "step": 2833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612959502148989e-06,
+      "loss": 2.37,
+      "step": 2834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61262363612352e-06,
+      "loss": 2.4397,
+      "step": 2835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612287630304911e-06,
+      "loss": 2.2737,
+      "step": 2836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611951484703347e-06,
+      "loss": 2.381,
+      "step": 2837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611615199329013e-06,
+      "loss": 2.5233,
+      "step": 2838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611278774192102e-06,
+      "loss": 2.3168,
+      "step": 2839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61094220930281e-06,
+      "loss": 2.3835,
+      "step": 2840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610605504671338e-06,
+      "loss": 2.3973,
+      "step": 2841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610268660307886e-06,
+      "loss": 2.3229,
+      "step": 2842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609931676222669e-06,
+      "loss": 2.5527,
+      "step": 2843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609594552425897e-06,
+      "loss": 2.8364,
+      "step": 2844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609257288927785e-06,
+      "loss": 2.473,
+      "step": 2845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608919885738557e-06,
+      "loss": 2.3863,
+      "step": 2846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608582342868438e-06,
+      "loss": 2.3991,
+      "step": 2847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608244660327657e-06,
+      "loss": 2.5306,
+      "step": 2848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607906838126451e-06,
+      "loss": 2.1877,
+      "step": 2849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607568876275052e-06,
+      "loss": 2.6217,
+      "step": 2850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607230774783709e-06,
+      "loss": 2.3765,
+      "step": 2851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606892533662665e-06,
+      "loss": 2.4494,
+      "step": 2852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606554152922173e-06,
+      "loss": 2.3978,
+      "step": 2853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606215632572486e-06,
+      "loss": 2.4322,
+      "step": 2854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605876972623866e-06,
+      "loss": 2.3227,
+      "step": 2855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605538173086575e-06,
+      "loss": 2.4759,
+      "step": 2856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60519923397088e-06,
+      "loss": 2.7272,
+      "step": 2857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604860155287054e-06,
+      "loss": 2.278,
+      "step": 2858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604520937045376e-06,
+      "loss": 2.3787,
+      "step": 2859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604181579256122e-06,
+      "loss": 2.541,
+      "step": 2860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60384208192958e-06,
+      "loss": 2.4069,
+      "step": 2861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603502445076038e-06,
+      "loss": 2.3401,
+      "step": 2862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603162668705787e-06,
+      "loss": 2.4299,
+      "step": 2863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60282275282913e-06,
+      "loss": 2.2574,
+      "step": 2864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602482697456364e-06,
+      "loss": 2.52,
+      "step": 2865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602142502597796e-06,
+      "loss": 2.5387,
+      "step": 2866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601802168263736e-06,
+      "loss": 2.2541,
+      "step": 2867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6014616944645e-06,
+      "loss": 2.1508,
+      "step": 2868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601121081210405e-06,
+      "loss": 2.3597,
+      "step": 2869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600780328511776e-06,
+      "loss": 2.5232,
+      "step": 2870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600439436378937e-06,
+      "loss": 2.4396,
+      "step": 2871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60009840482222e-06,
+      "loss": 2.6068,
+      "step": 2872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599757233851963e-06,
+      "loss": 2.332,
+      "step": 2873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599415923478502e-06,
+      "loss": 2.7064,
+      "step": 2874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599074473712185e-06,
+      "loss": 2.4584,
+      "step": 2875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598732884563357e-06,
+      "loss": 2.0659,
+      "step": 2876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598391156042371e-06,
+      "loss": 2.4715,
+      "step": 2877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598049288159585e-06,
+      "loss": 2.2735,
+      "step": 2878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59770728092536e-06,
+      "loss": 2.4684,
+      "step": 2879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597365134350059e-06,
+      "loss": 2.3911,
+      "step": 2880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597022848444053e-06,
+      "loss": 2.3876,
+      "step": 2881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596680423217714e-06,
+      "loss": 2.2478,
+      "step": 2882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596337858681423e-06,
+      "loss": 2.27,
+      "step": 2883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595995154845558e-06,
+      "loss": 2.3891,
+      "step": 2884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595652311720509e-06,
+      "loss": 2.3702,
+      "step": 2885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595309329316663e-06,
+      "loss": 2.4401,
+      "step": 2886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594966207644415e-06,
+      "loss": 2.203,
+      "step": 2887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594622946714166e-06,
+      "loss": 2.4569,
+      "step": 2888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594279546536318e-06,
+      "loss": 2.7382,
+      "step": 2889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593936007121279e-06,
+      "loss": 2.558,
+      "step": 2890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593592328479459e-06,
+      "loss": 2.3184,
+      "step": 2891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593248510621274e-06,
+      "loss": 2.2807,
+      "step": 2892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592904553557146e-06,
+      "loss": 2.411,
+      "step": 2893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592560457297497e-06,
+      "loss": 2.5049,
+      "step": 2894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592216221852757e-06,
+      "loss": 2.2418,
+      "step": 2895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591871847233356e-06,
+      "loss": 2.1307,
+      "step": 2896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591527333449734e-06,
+      "loss": 2.6007,
+      "step": 2897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59118268051233e-06,
+      "loss": 2.5991,
+      "step": 2898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59083788843159e-06,
+      "loss": 2.3547,
+      "step": 2899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590492957217963e-06,
+      "loss": 2.2425,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590147886881904e-06,
+      "loss": 2.6021,
+      "step": 2901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589802677433871e-06,
+      "loss": 2.3873,
+      "step": 2902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589457328884324e-06,
+      "loss": 2.2381,
+      "step": 2903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58911184124373e-06,
+      "loss": 2.4584,
+      "step": 2904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588766214522562e-06,
+      "loss": 2.4956,
+      "step": 2905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58842044873129e-06,
+      "loss": 2.3988,
+      "step": 2906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588074543880399e-06,
+      "loss": 2.2188,
+      "step": 2907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587728499980367e-06,
+      "loss": 2.549,
+      "step": 2908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587382317041684e-06,
+      "loss": 2.2748,
+      "step": 2909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587035995074842e-06,
+      "loss": 2.5544,
+      "step": 2910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586689534090336e-06,
+      "loss": 2.1337,
+      "step": 2911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586342934098664e-06,
+      "loss": 2.2695,
+      "step": 2912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585996195110335e-06,
+      "loss": 2.2029,
+      "step": 2913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585649317135855e-06,
+      "loss": 2.4408,
+      "step": 2914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585302300185736e-06,
+      "loss": 2.4986,
+      "step": 2915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584955144270493e-06,
+      "loss": 2.242,
+      "step": 2916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584607849400652e-06,
+      "loss": 2.0249,
+      "step": 2917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584260415586737e-06,
+      "loss": 2.4908,
+      "step": 2918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583912842839275e-06,
+      "loss": 2.3395,
+      "step": 2919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583565131168802e-06,
+      "loss": 2.4516,
+      "step": 2920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583217280585855e-06,
+      "loss": 2.626,
+      "step": 2921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582869291100976e-06,
+      "loss": 2.5213,
+      "step": 2922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582521162724712e-06,
+      "loss": 2.1948,
+      "step": 2923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582172895467614e-06,
+      "loss": 2.3119,
+      "step": 2924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581824489340235e-06,
+      "loss": 2.3538,
+      "step": 2925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581475944353136e-06,
+      "loss": 2.4118,
+      "step": 2926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58112726051688e-06,
+      "loss": 2.3714,
+      "step": 2927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580778437842035e-06,
+      "loss": 2.5933,
+      "step": 2928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580429476339171e-06,
+      "loss": 2.3241,
+      "step": 2929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580080376018863e-06,
+      "loss": 2.5025,
+      "step": 2930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579731136891694e-06,
+      "loss": 2.693,
+      "step": 2931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579381758968244e-06,
+      "loss": 2.5223,
+      "step": 2932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579032242259107e-06,
+      "loss": 2.2937,
+      "step": 2933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578682586774874e-06,
+      "loss": 2.4718,
+      "step": 2934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578332792526139e-06,
+      "loss": 2.1796,
+      "step": 2935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577982859523505e-06,
+      "loss": 2.3114,
+      "step": 2936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577632787777577e-06,
+      "loss": 2.2048,
+      "step": 2937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577282577298966e-06,
+      "loss": 2.246,
+      "step": 2938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576932228098282e-06,
+      "loss": 2.5229,
+      "step": 2939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576581740186147e-06,
+      "loss": 2.3918,
+      "step": 2940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57623111357318e-06,
+      "loss": 2.2571,
+      "step": 2941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575880348270012e-06,
+      "loss": 2.4832,
+      "step": 2942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575529444287268e-06,
+      "loss": 2.2438,
+      "step": 2943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575178401635586e-06,
+      "loss": 2.5234,
+      "step": 2944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574827220325601e-06,
+      "loss": 2.5823,
+      "step": 2945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574475900367962e-06,
+      "loss": 2.452,
+      "step": 2946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574124441773313e-06,
+      "loss": 2.369,
+      "step": 2947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573772844552304e-06,
+      "loss": 2.4896,
+      "step": 2948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573421108715594e-06,
+      "loss": 2.3999,
+      "step": 2949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573069234273841e-06,
+      "loss": 2.3204,
+      "step": 2950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572717221237712e-06,
+      "loss": 2.7006,
+      "step": 2951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572365069617871e-06,
+      "loss": 2.3874,
+      "step": 2952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572012779424994e-06,
+      "loss": 2.7425,
+      "step": 2953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571660350669754e-06,
+      "loss": 2.3452,
+      "step": 2954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571307783362836e-06,
+      "loss": 2.3083,
+      "step": 2955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570955077514923e-06,
+      "loss": 2.6986,
+      "step": 2956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570602233136703e-06,
+      "loss": 2.4563,
+      "step": 2957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570249250238875e-06,
+      "loss": 2.2593,
+      "step": 2958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56989612883213e-06,
+      "loss": 2.3388,
+      "step": 2959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569542868927174e-06,
+      "loss": 2.1999,
+      "step": 2960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56918947053471e-06,
+      "loss": 2.3221,
+      "step": 2961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568835933665452e-06,
+      "loss": 2.6184,
+      "step": 2962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568482258330113e-06,
+      "loss": 2.6899,
+      "step": 2963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56812844453941e-06,
+      "loss": 2.1732,
+      "step": 2964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567774492304069e-06,
+      "loss": 2.4226,
+      "step": 2965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567420401634816e-06,
+      "loss": 2.6427,
+      "step": 2966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56706617254238e-06,
+      "loss": 2.1164,
+      "step": 2967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5667118050375e-06,
+      "loss": 2.3723,
+      "step": 2968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566357299130913e-06,
+      "loss": 2.5499,
+      "step": 2969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566002654833364e-06,
+      "loss": 2.3809,
+      "step": 2970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565647872155603e-06,
+      "loss": 2.2885,
+      "step": 2971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565292951108378e-06,
+      "loss": 2.258,
+      "step": 2972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564937891702448e-06,
+      "loss": 2.368,
+      "step": 2973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564582693948575e-06,
+      "loss": 2.1054,
+      "step": 2974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56422735785752e-06,
+      "loss": 2.4287,
+      "step": 2975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563871883440054e-06,
+      "loss": 2.5579,
+      "step": 2976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563516270706952e-06,
+      "loss": 2.166,
+      "step": 2977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56316051966899e-06,
+      "loss": 2.5041,
+      "step": 2978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56280463033695e-06,
+      "loss": 2.4699,
+      "step": 2979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562448602721616e-06,
+      "loss": 2.3111,
+      "step": 2980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56209243683378e-06,
+      "loss": 1.9897,
+      "step": 2981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561736132684235e-06,
+      "loss": 2.2252,
+      "step": 2982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561379690283782e-06,
+      "loss": 2.5421,
+      "step": 2983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56102310964322e-06,
+      "loss": 2.3278,
+      "step": 2984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560666390773356e-06,
+      "loss": 2.2893,
+      "step": 2985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560309533685004e-06,
+      "loss": 2.4516,
+      "step": 2986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559952538388978e-06,
+      "loss": 2.3989,
+      "step": 2987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559595404896096e-06,
+      "loss": 2.2577,
+      "step": 2988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559238133217182e-06,
+      "loss": 2.6105,
+      "step": 2989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558880723363063e-06,
+      "loss": 2.2453,
+      "step": 2990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558523175344575e-06,
+      "loss": 2.5858,
+      "step": 2991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558165489172547e-06,
+      "loss": 2.0544,
+      "step": 2992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557807664857825e-06,
+      "loss": 2.363,
+      "step": 2993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557449702411252e-06,
+      "loss": 2.4337,
+      "step": 2994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557091601843676e-06,
+      "loss": 2.4868,
+      "step": 2995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55673336316595e-06,
+      "loss": 2.185,
+      "step": 2996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55637498638893e-06,
+      "loss": 2.4583,
+      "step": 2997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556016471523479e-06,
+      "loss": 2.568,
+      "step": 2998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555657818580463e-06,
+      "loss": 2.41,
+      "step": 2999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55529902757075e-06,
+      "loss": 2.1368,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 18605,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.575273271243047e+18,
+  "trial_name": null,
+  "trial_params": null
+}