End of training

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.0,
+    "train_loss": 0.8650754356088861,
+    "train_runtime": 3796.8514,
+    "train_samples": 87599,
+    "train_samples_per_second": 46.143,
+    "train_steps_per_second": 0.181
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 2.0,
+    "train_loss": 0.8650754356088861,
+    "train_runtime": 3796.8514,
+    "train_samples": 87599,
+    "train_samples_per_second": 46.143,
+    "train_steps_per_second": 0.181
+}

trainer_state.json

@@ -0,0 +1,4141 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 686,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9956268221574345e-05,
+      "loss": 5.3629,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.991253644314869e-05,
+      "loss": 3.9386,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9868804664723033e-05,
+      "loss": 3.4456,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9825072886297377e-05,
+      "loss": 3.3224,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.978134110787172e-05,
+      "loss": 2.9937,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9737609329446064e-05,
+      "loss": 2.8936,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.969387755102041e-05,
+      "loss": 2.6102,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9650145772594756e-05,
+      "loss": 2.5955,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9606413994169096e-05,
+      "loss": 2.2454,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.956268221574344e-05,
+      "loss": 2.294,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9518950437317784e-05,
+      "loss": 2.01,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9475218658892128e-05,
+      "loss": 2.1138,
+      "step": 12
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9431486880466475e-05,
+      "loss": 1.8897,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9387755102040816e-05,
+      "loss": 1.8003,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9344023323615163e-05,
+      "loss": 1.7822,
+      "step": 15
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9300291545189504e-05,
+      "loss": 1.8079,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.925655976676385e-05,
+      "loss": 1.9259,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9212827988338192e-05,
+      "loss": 1.5808,
+      "step": 18
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9169096209912536e-05,
+      "loss": 1.621,
+      "step": 19
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9125364431486883e-05,
+      "loss": 1.651,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9081632653061224e-05,
+      "loss": 1.5975,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.903790087463557e-05,
+      "loss": 1.5772,
+      "step": 22
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.899416909620991e-05,
+      "loss": 1.7805,
+      "step": 23
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.895043731778426e-05,
+      "loss": 1.5725,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.89067055393586e-05,
+      "loss": 1.4716,
+      "step": 25
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8862973760932947e-05,
+      "loss": 1.6262,
+      "step": 26
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.881924198250729e-05,
+      "loss": 1.536,
+      "step": 27
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.877551020408163e-05,
+      "loss": 1.5282,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.873177842565598e-05,
+      "loss": 1.506,
+      "step": 29
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.868804664723032e-05,
+      "loss": 1.4047,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8644314868804667e-05,
+      "loss": 1.5451,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.860058309037901e-05,
+      "loss": 1.2876,
+      "step": 32
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8556851311953354e-05,
+      "loss": 1.3083,
+      "step": 33
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.85131195335277e-05,
+      "loss": 1.3228,
+      "step": 34
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8469387755102042e-05,
+      "loss": 1.2497,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8425655976676386e-05,
+      "loss": 1.2854,
+      "step": 36
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8381924198250727e-05,
+      "loss": 1.332,
+      "step": 37
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8338192419825074e-05,
+      "loss": 1.2977,
+      "step": 38
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8294460641399418e-05,
+      "loss": 1.1913,
+      "step": 39
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8250728862973762e-05,
+      "loss": 1.4206,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8206997084548106e-05,
+      "loss": 1.3032,
+      "step": 41
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 1.1998,
+      "step": 42
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8119533527696794e-05,
+      "loss": 1.1315,
+      "step": 43
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8075801749271134e-05,
+      "loss": 1.199,
+      "step": 44
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8032069970845482e-05,
+      "loss": 1.1658,
+      "step": 45
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.7988338192419826e-05,
+      "loss": 1.1449,
+      "step": 46
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.794460641399417e-05,
+      "loss": 1.1687,
+      "step": 47
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7900874635568514e-05,
+      "loss": 1.2565,
+      "step": 48
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7857142857142858e-05,
+      "loss": 1.1276,
+      "step": 49
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.78134110787172e-05,
+      "loss": 1.1296,
+      "step": 50
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.776967930029155e-05,
+      "loss": 1.2696,
+      "step": 51
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.772594752186589e-05,
+      "loss": 1.3116,
+      "step": 52
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7682215743440233e-05,
+      "loss": 1.2603,
+      "step": 53
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7638483965014577e-05,
+      "loss": 1.1573,
+      "step": 54
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.759475218658892e-05,
+      "loss": 1.1059,
+      "step": 55
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 1.0841,
+      "step": 56
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.750728862973761e-05,
+      "loss": 1.1224,
+      "step": 57
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7463556851311957e-05,
+      "loss": 1.0036,
+      "step": 58
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7419825072886297e-05,
+      "loss": 1.1716,
+      "step": 59
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7376093294460644e-05,
+      "loss": 1.1219,
+      "step": 60
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7332361516034985e-05,
+      "loss": 1.3287,
+      "step": 61
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.728862973760933e-05,
+      "loss": 1.1589,
+      "step": 62
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7244897959183673e-05,
+      "loss": 1.1281,
+      "step": 63
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7201166180758017e-05,
+      "loss": 1.0111,
+      "step": 64
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7157434402332364e-05,
+      "loss": 1.0364,
+      "step": 65
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7113702623906705e-05,
+      "loss": 0.9568,
+      "step": 66
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7069970845481052e-05,
+      "loss": 1.0818,
+      "step": 67
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7026239067055393e-05,
+      "loss": 1.3587,
+      "step": 68
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.698250728862974e-05,
+      "loss": 1.0199,
+      "step": 69
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 1.1711,
+      "step": 70
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6895043731778424e-05,
+      "loss": 1.0558,
+      "step": 71
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6851311953352772e-05,
+      "loss": 1.1889,
+      "step": 72
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6807580174927112e-05,
+      "loss": 0.8719,
+      "step": 73
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.676384839650146e-05,
+      "loss": 1.0826,
+      "step": 74
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.67201166180758e-05,
+      "loss": 0.8779,
+      "step": 75
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6676384839650148e-05,
+      "loss": 1.0484,
+      "step": 76
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.663265306122449e-05,
+      "loss": 1.2392,
+      "step": 77
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6588921282798835e-05,
+      "loss": 1.0915,
+      "step": 78
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.654518950437318e-05,
+      "loss": 0.9508,
+      "step": 79
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.650145772594752e-05,
+      "loss": 1.0686,
+      "step": 80
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6457725947521867e-05,
+      "loss": 1.0613,
+      "step": 81
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6413994169096208e-05,
+      "loss": 0.9126,
+      "step": 82
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6370262390670555e-05,
+      "loss": 1.2267,
+      "step": 83
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.63265306122449e-05,
+      "loss": 0.9892,
+      "step": 84
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6282798833819243e-05,
+      "loss": 0.9648,
+      "step": 85
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6239067055393587e-05,
+      "loss": 0.9477,
+      "step": 86
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.619533527696793e-05,
+      "loss": 1.0255,
+      "step": 87
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6151603498542275e-05,
+      "loss": 0.9431,
+      "step": 88
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.610787172011662e-05,
+      "loss": 0.9383,
+      "step": 89
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6064139941690963e-05,
+      "loss": 1.0558,
+      "step": 90
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6020408163265307e-05,
+      "loss": 0.9906,
+      "step": 91
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.597667638483965e-05,
+      "loss": 1.0479,
+      "step": 92
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5932944606413995e-05,
+      "loss": 1.0124,
+      "step": 93
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.588921282798834e-05,
+      "loss": 0.8342,
+      "step": 94
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5845481049562683e-05,
+      "loss": 1.0599,
+      "step": 95
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.580174927113703e-05,
+      "loss": 1.0933,
+      "step": 96
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.575801749271137e-05,
+      "loss": 1.0057,
+      "step": 97
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 1.0297,
+      "step": 98
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.567055393586006e-05,
+      "loss": 1.1918,
+      "step": 99
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5626822157434402e-05,
+      "loss": 0.9445,
+      "step": 100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5583090379008746e-05,
+      "loss": 1.1254,
+      "step": 101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.553935860058309e-05,
+      "loss": 0.9527,
+      "step": 102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5495626822157438e-05,
+      "loss": 0.8619,
+      "step": 103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5451895043731778e-05,
+      "loss": 0.9988,
+      "step": 104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5408163265306125e-05,
+      "loss": 0.7942,
+      "step": 105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5364431486880466e-05,
+      "loss": 1.0989,
+      "step": 106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.532069970845481e-05,
+      "loss": 0.9513,
+      "step": 107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5276967930029154e-05,
+      "loss": 1.0775,
+      "step": 108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5233236151603498e-05,
+      "loss": 1.1696,
+      "step": 109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5189504373177845e-05,
+      "loss": 1.0216,
+      "step": 110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5145772594752186e-05,
+      "loss": 1.03,
+      "step": 111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5102040816326533e-05,
+      "loss": 0.9126,
+      "step": 112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5058309037900874e-05,
+      "loss": 0.8499,
+      "step": 113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.501457725947522e-05,
+      "loss": 0.8761,
+      "step": 114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4970845481049565e-05,
+      "loss": 1.0023,
+      "step": 115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4927113702623906e-05,
+      "loss": 0.9489,
+      "step": 116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4883381924198253e-05,
+      "loss": 0.9472,
+      "step": 117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4839650145772593e-05,
+      "loss": 0.9639,
+      "step": 118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.479591836734694e-05,
+      "loss": 1.0146,
+      "step": 119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.475218658892128e-05,
+      "loss": 1.0069,
+      "step": 120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470845481049563e-05,
+      "loss": 1.0586,
+      "step": 121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4664723032069973e-05,
+      "loss": 1.2354,
+      "step": 122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4620991253644317e-05,
+      "loss": 0.9995,
+      "step": 123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.457725947521866e-05,
+      "loss": 0.9738,
+      "step": 124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4533527696793e-05,
+      "loss": 1.0605,
+      "step": 125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 0.9859,
+      "step": 126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4446064139941692e-05,
+      "loss": 1.0085,
+      "step": 127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4402332361516036e-05,
+      "loss": 0.8942,
+      "step": 128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.435860058309038e-05,
+      "loss": 0.9521,
+      "step": 129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4314868804664724e-05,
+      "loss": 0.9921,
+      "step": 130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4271137026239068e-05,
+      "loss": 0.985,
+      "step": 131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.422740524781341e-05,
+      "loss": 0.8275,
+      "step": 132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4183673469387756e-05,
+      "loss": 1.0239,
+      "step": 133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.41399416909621e-05,
+      "loss": 0.9122,
+      "step": 134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4096209912536444e-05,
+      "loss": 0.912,
+      "step": 135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4052478134110788e-05,
+      "loss": 0.9607,
+      "step": 136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4008746355685132e-05,
+      "loss": 1.101,
+      "step": 137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3965014577259476e-05,
+      "loss": 0.9501,
+      "step": 138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.392128279883382e-05,
+      "loss": 0.8686,
+      "step": 139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3877551020408164e-05,
+      "loss": 0.9697,
+      "step": 140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3833819241982508e-05,
+      "loss": 0.9529,
+      "step": 141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.379008746355685e-05,
+      "loss": 1.0023,
+      "step": 142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3746355685131196e-05,
+      "loss": 0.9224,
+      "step": 143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.370262390670554e-05,
+      "loss": 0.8481,
+      "step": 144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3658892128279883e-05,
+      "loss": 0.8655,
+      "step": 145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3615160349854227e-05,
+      "loss": 0.9538,
+      "step": 146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.357142857142857e-05,
+      "loss": 0.9949,
+      "step": 147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.352769679300292e-05,
+      "loss": 0.9171,
+      "step": 148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.348396501457726e-05,
+      "loss": 0.8127,
+      "step": 149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3440233236151603e-05,
+      "loss": 0.9542,
+      "step": 150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3396501457725947e-05,
+      "loss": 0.8685,
+      "step": 151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.335276967930029e-05,
+      "loss": 1.0402,
+      "step": 152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.330903790087464e-05,
+      "loss": 0.978,
+      "step": 153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.326530612244898e-05,
+      "loss": 0.7553,
+      "step": 154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3221574344023326e-05,
+      "loss": 0.9466,
+      "step": 155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3177842565597667e-05,
+      "loss": 0.9465,
+      "step": 156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3134110787172014e-05,
+      "loss": 0.8261,
+      "step": 157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3090379008746355e-05,
+      "loss": 0.957,
+      "step": 158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.30466472303207e-05,
+      "loss": 0.7844,
+      "step": 159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3002915451895046e-05,
+      "loss": 0.8258,
+      "step": 160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2959183673469387e-05,
+      "loss": 1.0157,
+      "step": 161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2915451895043734e-05,
+      "loss": 0.9961,
+      "step": 162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2871720116618074e-05,
+      "loss": 0.8449,
+      "step": 163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2827988338192422e-05,
+      "loss": 0.8206,
+      "step": 164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2784256559766762e-05,
+      "loss": 0.815,
+      "step": 165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.274052478134111e-05,
+      "loss": 0.9007,
+      "step": 166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2696793002915454e-05,
+      "loss": 0.8002,
+      "step": 167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2653061224489794e-05,
+      "loss": 0.8584,
+      "step": 168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.260932944606414e-05,
+      "loss": 0.8984,
+      "step": 169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2565597667638482e-05,
+      "loss": 0.9296,
+      "step": 170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.252186588921283e-05,
+      "loss": 0.9471,
+      "step": 171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2478134110787173e-05,
+      "loss": 0.9958,
+      "step": 172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2434402332361517e-05,
+      "loss": 0.9842,
+      "step": 173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.239067055393586e-05,
+      "loss": 0.9085,
+      "step": 174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2346938775510205e-05,
+      "loss": 0.8902,
+      "step": 175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.230320699708455e-05,
+      "loss": 1.108,
+      "step": 176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.225947521865889e-05,
+      "loss": 0.8656,
+      "step": 177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2215743440233237e-05,
+      "loss": 0.9684,
+      "step": 178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.217201166180758e-05,
+      "loss": 1.0054,
+      "step": 179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2128279883381925e-05,
+      "loss": 0.9552,
+      "step": 180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.208454810495627e-05,
+      "loss": 0.7809,
+      "step": 181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 1.0736,
+      "step": 182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1997084548104957e-05,
+      "loss": 0.8351,
+      "step": 183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.19533527696793e-05,
+      "loss": 1.0844,
+      "step": 184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1909620991253645e-05,
+      "loss": 0.8503,
+      "step": 185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.186588921282799e-05,
+      "loss": 0.9959,
+      "step": 186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1822157434402333e-05,
+      "loss": 1.0989,
+      "step": 187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1778425655976677e-05,
+      "loss": 0.9763,
+      "step": 188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.173469387755102e-05,
+      "loss": 1.0347,
+      "step": 189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1690962099125364e-05,
+      "loss": 0.9371,
+      "step": 190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1647230320699712e-05,
+      "loss": 0.7817,
+      "step": 191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1603498542274052e-05,
+      "loss": 0.9467,
+      "step": 192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.15597667638484e-05,
+      "loss": 0.9691,
+      "step": 193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.151603498542274e-05,
+      "loss": 0.7149,
+      "step": 194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1472303206997084e-05,
+      "loss": 0.7316,
+      "step": 195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.7645,
+      "step": 196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1384839650145772e-05,
+      "loss": 0.7922,
+      "step": 197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.134110787172012e-05,
+      "loss": 0.9184,
+      "step": 198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.129737609329446e-05,
+      "loss": 0.8611,
+      "step": 199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1253644314868807e-05,
+      "loss": 0.8158,
+      "step": 200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1209912536443148e-05,
+      "loss": 1.0274,
+      "step": 201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1166180758017495e-05,
+      "loss": 0.9442,
+      "step": 202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1122448979591836e-05,
+      "loss": 1.1595,
+      "step": 203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.107871720116618e-05,
+      "loss": 0.8674,
+      "step": 204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1034985422740527e-05,
+      "loss": 0.9162,
+      "step": 205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0991253644314868e-05,
+      "loss": 0.8482,
+      "step": 206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0947521865889215e-05,
+      "loss": 0.8585,
+      "step": 207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0903790087463556e-05,
+      "loss": 0.9633,
+      "step": 208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0860058309037903e-05,
+      "loss": 0.9125,
+      "step": 209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 0.7642,
+      "step": 210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.077259475218659e-05,
+      "loss": 0.8236,
+      "step": 211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0728862973760935e-05,
+      "loss": 0.7912,
+      "step": 212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0685131195335275e-05,
+      "loss": 0.8434,
+      "step": 213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0641399416909623e-05,
+      "loss": 0.8584,
+      "step": 214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0597667638483963e-05,
+      "loss": 0.8301,
+      "step": 215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.055393586005831e-05,
+      "loss": 0.8389,
+      "step": 216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0510204081632654e-05,
+      "loss": 0.9902,
+      "step": 217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0466472303207e-05,
+      "loss": 0.8553,
+      "step": 218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0422740524781342e-05,
+      "loss": 0.872,
+      "step": 219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0379008746355683e-05,
+      "loss": 0.8447,
+      "step": 220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.033527696793003e-05,
+      "loss": 0.8772,
+      "step": 221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.029154518950437e-05,
+      "loss": 0.8157,
+      "step": 222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0247813411078718e-05,
+      "loss": 0.7805,
+      "step": 223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0204081632653062e-05,
+      "loss": 0.708,
+      "step": 224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0160349854227406e-05,
+      "loss": 0.7722,
+      "step": 225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.011661807580175e-05,
+      "loss": 0.9642,
+      "step": 226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0072886297376094e-05,
+      "loss": 0.7912,
+      "step": 227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0029154518950438e-05,
+      "loss": 1.0257,
+      "step": 228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9985422740524782e-05,
+      "loss": 0.9954,
+      "step": 229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9941690962099126e-05,
+      "loss": 0.9803,
+      "step": 230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.989795918367347e-05,
+      "loss": 0.7859,
+      "step": 231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9854227405247814e-05,
+      "loss": 0.8381,
+      "step": 232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9810495626822158e-05,
+      "loss": 0.6901,
+      "step": 233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.97667638483965e-05,
+      "loss": 1.0198,
+      "step": 234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9723032069970846e-05,
+      "loss": 0.9632,
+      "step": 235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9679300291545193e-05,
+      "loss": 0.7811,
+      "step": 236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9635568513119533e-05,
+      "loss": 0.776,
+      "step": 237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.7588,
+      "step": 238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.954810495626822e-05,
+      "loss": 0.8946,
+      "step": 239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9504373177842565e-05,
+      "loss": 0.8895,
+      "step": 240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.946064139941691e-05,
+      "loss": 0.9084,
+      "step": 241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9416909620991253e-05,
+      "loss": 0.8147,
+      "step": 242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93731778425656e-05,
+      "loss": 0.8603,
+      "step": 243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.932944606413994e-05,
+      "loss": 0.8547,
+      "step": 244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.7791,
+      "step": 245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924198250728863e-05,
+      "loss": 0.8709,
+      "step": 246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9198250728862973e-05,
+      "loss": 0.7485,
+      "step": 247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.915451895043732e-05,
+      "loss": 0.8703,
+      "step": 248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.911078717201166e-05,
+      "loss": 0.9068,
+      "step": 249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9067055393586008e-05,
+      "loss": 0.8725,
+      "step": 250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.902332361516035e-05,
+      "loss": 0.7559,
+      "step": 251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8979591836734696e-05,
+      "loss": 0.8044,
+      "step": 252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8935860058309037e-05,
+      "loss": 1.0526,
+      "step": 253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8892128279883384e-05,
+      "loss": 0.8945,
+      "step": 254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8848396501457728e-05,
+      "loss": 0.799,
+      "step": 255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.880466472303207e-05,
+      "loss": 0.7597,
+      "step": 256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8760932944606416e-05,
+      "loss": 0.772,
+      "step": 257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8717201166180756e-05,
+      "loss": 0.9276,
+      "step": 258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8673469387755104e-05,
+      "loss": 0.7726,
+      "step": 259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8629737609329444e-05,
+      "loss": 0.9847,
+      "step": 260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.858600583090379e-05,
+      "loss": 0.8455,
+      "step": 261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8542274052478135e-05,
+      "loss": 0.7554,
+      "step": 262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.849854227405248e-05,
+      "loss": 0.7653,
+      "step": 263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8454810495626823e-05,
+      "loss": 0.8519,
+      "step": 264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8411078717201164e-05,
+      "loss": 0.7033,
+      "step": 265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.8955,
+      "step": 266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8323615160349855e-05,
+      "loss": 0.7778,
+      "step": 267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.82798833819242e-05,
+      "loss": 0.819,
+      "step": 268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8236151603498543e-05,
+      "loss": 0.8427,
+      "step": 269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8192419825072887e-05,
+      "loss": 0.8098,
+      "step": 270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.814868804664723e-05,
+      "loss": 0.8787,
+      "step": 271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8104956268221575e-05,
+      "loss": 0.7492,
+      "step": 272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.806122448979592e-05,
+      "loss": 0.7385,
+      "step": 273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8017492711370263e-05,
+      "loss": 0.7456,
+      "step": 274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7973760932944607e-05,
+      "loss": 0.8629,
+      "step": 275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.793002915451895e-05,
+      "loss": 0.9145,
+      "step": 276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7886297376093295e-05,
+      "loss": 0.8599,
+      "step": 277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784256559766764e-05,
+      "loss": 0.8948,
+      "step": 278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7798833819241983e-05,
+      "loss": 0.8635,
+      "step": 279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7755102040816327e-05,
+      "loss": 0.7858,
+      "step": 280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7711370262390674e-05,
+      "loss": 0.9611,
+      "step": 281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7667638483965014e-05,
+      "loss": 0.814,
+      "step": 282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.762390670553936e-05,
+      "loss": 0.8337,
+      "step": 283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7580174927113702e-05,
+      "loss": 0.7874,
+      "step": 284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7536443148688046e-05,
+      "loss": 0.7569,
+      "step": 285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7492711370262394e-05,
+      "loss": 0.8028,
+      "step": 286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7448979591836734e-05,
+      "loss": 0.7952,
+      "step": 287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.740524781341108e-05,
+      "loss": 0.8058,
+      "step": 288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7361516034985422e-05,
+      "loss": 0.7663,
+      "step": 289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.731778425655977e-05,
+      "loss": 0.8116,
+      "step": 290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.727405247813411e-05,
+      "loss": 0.8715,
+      "step": 291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7230320699708454e-05,
+      "loss": 0.9337,
+      "step": 292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.71865889212828e-05,
+      "loss": 0.6941,
+      "step": 293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.7186,
+      "step": 294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.709912536443149e-05,
+      "loss": 0.8248,
+      "step": 295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.705539358600583e-05,
+      "loss": 0.9292,
+      "step": 296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7011661807580177e-05,
+      "loss": 0.8749,
+      "step": 297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6967930029154518e-05,
+      "loss": 0.9838,
+      "step": 298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6924198250728865e-05,
+      "loss": 0.8464,
+      "step": 299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.688046647230321e-05,
+      "loss": 0.8021,
+      "step": 300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.683673469387755e-05,
+      "loss": 0.8395,
+      "step": 301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6793002915451897e-05,
+      "loss": 0.7987,
+      "step": 302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6749271137026237e-05,
+      "loss": 0.8711,
+      "step": 303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6705539358600585e-05,
+      "loss": 0.752,
+      "step": 304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.666180758017493e-05,
+      "loss": 0.7489,
+      "step": 305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6618075801749273e-05,
+      "loss": 0.8933,
+      "step": 306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6574344023323617e-05,
+      "loss": 0.8087,
+      "step": 307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6530612244897957e-05,
+      "loss": 0.9219,
+      "step": 308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6486880466472304e-05,
+      "loss": 0.7434,
+      "step": 309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443148688046645e-05,
+      "loss": 0.6875,
+      "step": 310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6399416909620992e-05,
+      "loss": 0.8969,
+      "step": 311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355685131195336e-05,
+      "loss": 0.7929,
+      "step": 312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631195335276968e-05,
+      "loss": 0.783,
+      "step": 313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6268221574344024e-05,
+      "loss": 0.98,
+      "step": 314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6224489795918368e-05,
+      "loss": 0.8074,
+      "step": 315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6180758017492712e-05,
+      "loss": 0.8186,
+      "step": 316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6137026239067053e-05,
+      "loss": 0.6993,
+      "step": 317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.60932944606414e-05,
+      "loss": 0.8308,
+      "step": 318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6049562682215744e-05,
+      "loss": 0.7593,
+      "step": 319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6005830903790088e-05,
+      "loss": 0.6809,
+      "step": 320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5962099125364432e-05,
+      "loss": 0.7695,
+      "step": 321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.917,
+      "step": 322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.587463556851312e-05,
+      "loss": 0.7101,
+      "step": 323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5830903790087464e-05,
+      "loss": 0.9573,
+      "step": 324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5787172011661808e-05,
+      "loss": 0.8902,
+      "step": 325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.574344023323615e-05,
+      "loss": 0.7231,
+      "step": 326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5699708454810496e-05,
+      "loss": 0.8754,
+      "step": 327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.565597667638484e-05,
+      "loss": 0.9213,
+      "step": 328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5612244897959183e-05,
+      "loss": 0.7397,
+      "step": 329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5568513119533527e-05,
+      "loss": 0.7227,
+      "step": 330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5524781341107875e-05,
+      "loss": 0.7672,
+      "step": 331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5481049562682215e-05,
+      "loss": 0.7325,
+      "step": 332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5437317784256563e-05,
+      "loss": 0.8339,
+      "step": 333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5393586005830903e-05,
+      "loss": 0.793,
+      "step": 334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5349854227405247e-05,
+      "loss": 0.9217,
+      "step": 335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.530612244897959e-05,
+      "loss": 0.9395,
+      "step": 336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5262390670553935e-05,
+      "loss": 0.7461,
+      "step": 337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.521865889212828e-05,
+      "loss": 0.7148,
+      "step": 338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5174927113702623e-05,
+      "loss": 0.9149,
+      "step": 339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.513119533527697e-05,
+      "loss": 0.812,
+      "step": 340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5087463556851312e-05,
+      "loss": 0.6849,
+      "step": 341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5043731778425658e-05,
+      "loss": 0.7184,
+      "step": 342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5e-05,
+      "loss": 0.8343,
+      "step": 343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4956268221574344e-05,
+      "loss": 0.6949,
+      "step": 344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4912536443148688e-05,
+      "loss": 0.6325,
+      "step": 345
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4868804664723032e-05,
+      "loss": 0.6235,
+      "step": 346
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4825072886297378e-05,
+      "loss": 0.5982,
+      "step": 347
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.478134110787172e-05,
+      "loss": 0.803,
+      "step": 348
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4737609329446064e-05,
+      "loss": 0.6774,
+      "step": 349
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4693877551020408e-05,
+      "loss": 0.6396,
+      "step": 350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4650145772594752e-05,
+      "loss": 0.7721,
+      "step": 351
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4606413994169096e-05,
+      "loss": 0.8477,
+      "step": 352
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4562682215743442e-05,
+      "loss": 0.7166,
+      "step": 353
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4518950437317786e-05,
+      "loss": 0.6573,
+      "step": 354
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.447521865889213e-05,
+      "loss": 0.7047,
+      "step": 355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4431486880466473e-05,
+      "loss": 0.613,
+      "step": 356
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4387755102040816e-05,
+      "loss": 0.7446,
+      "step": 357
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.434402332361516e-05,
+      "loss": 0.668,
+      "step": 358
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4300291545189505e-05,
+      "loss": 0.6964,
+      "step": 359
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.425655976676385e-05,
+      "loss": 0.7486,
+      "step": 360
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4212827988338193e-05,
+      "loss": 0.7398,
+      "step": 361
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4169096209912537e-05,
+      "loss": 0.5944,
+      "step": 362
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4125364431486881e-05,
+      "loss": 0.6959,
+      "step": 363
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4081632653061225e-05,
+      "loss": 0.6982,
+      "step": 364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4037900874635567e-05,
+      "loss": 0.6906,
+      "step": 365
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3994169096209913e-05,
+      "loss": 0.6168,
+      "step": 366
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3950437317784257e-05,
+      "loss": 0.8569,
+      "step": 367
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.39067055393586e-05,
+      "loss": 0.696,
+      "step": 368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3862973760932945e-05,
+      "loss": 0.7874,
+      "step": 369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3819241982507289e-05,
+      "loss": 0.6416,
+      "step": 370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3775510204081633e-05,
+      "loss": 0.7262,
+      "step": 371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3731778425655978e-05,
+      "loss": 0.7177,
+      "step": 372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3688046647230322e-05,
+      "loss": 0.7477,
+      "step": 373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3644314868804664e-05,
+      "loss": 0.7449,
+      "step": 374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3600583090379008e-05,
+      "loss": 0.6565,
+      "step": 375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3556851311953352e-05,
+      "loss": 0.7586,
+      "step": 376
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3513119533527696e-05,
+      "loss": 0.519,
+      "step": 377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.711,
+      "step": 378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3425655976676386e-05,
+      "loss": 0.7435,
+      "step": 379
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.338192419825073e-05,
+      "loss": 0.822,
+      "step": 380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3338192419825074e-05,
+      "loss": 0.7531,
+      "step": 381
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3294460641399418e-05,
+      "loss": 0.8351,
+      "step": 382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.325072886297376e-05,
+      "loss": 0.6045,
+      "step": 383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3206997084548104e-05,
+      "loss": 0.8374,
+      "step": 384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.316326530612245e-05,
+      "loss": 0.7024,
+      "step": 385
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3119533527696794e-05,
+      "loss": 0.6995,
+      "step": 386
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3075801749271137e-05,
+      "loss": 0.7083,
+      "step": 387
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3032069970845481e-05,
+      "loss": 0.6404,
+      "step": 388
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2988338192419825e-05,
+      "loss": 0.8153,
+      "step": 389
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.294460641399417e-05,
+      "loss": 0.6466,
+      "step": 390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2900874635568515e-05,
+      "loss": 0.6655,
+      "step": 391
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.4714,
+      "step": 392
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2813411078717201e-05,
+      "loss": 0.7221,
+      "step": 393
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2769679300291545e-05,
+      "loss": 0.6697,
+      "step": 394
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2725947521865889e-05,
+      "loss": 0.8205,
+      "step": 395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2682215743440233e-05,
+      "loss": 0.6631,
+      "step": 396
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2638483965014577e-05,
+      "loss": 0.6889,
+      "step": 397
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2594752186588923e-05,
+      "loss": 0.7284,
+      "step": 398
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2551020408163267e-05,
+      "loss": 0.7834,
+      "step": 399
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.250728862973761e-05,
+      "loss": 0.6491,
+      "step": 400
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2463556851311953e-05,
+      "loss": 0.796,
+      "step": 401
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2419825072886297e-05,
+      "loss": 0.7083,
+      "step": 402
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.237609329446064e-05,
+      "loss": 0.6393,
+      "step": 403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2332361516034986e-05,
+      "loss": 0.7,
+      "step": 404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.228862973760933e-05,
+      "loss": 0.7005,
+      "step": 405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.635,
+      "step": 406
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2201166180758018e-05,
+      "loss": 0.7491,
+      "step": 407
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2157434402332362e-05,
+      "loss": 0.6133,
+      "step": 408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2113702623906704e-05,
+      "loss": 0.6128,
+      "step": 409
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.206997084548105e-05,
+      "loss": 0.6358,
+      "step": 410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2026239067055394e-05,
+      "loss": 0.8197,
+      "step": 411
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1982507288629738e-05,
+      "loss": 0.7669,
+      "step": 412
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1938775510204082e-05,
+      "loss": 0.6701,
+      "step": 413
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1895043731778426e-05,
+      "loss": 0.7437,
+      "step": 414
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.185131195335277e-05,
+      "loss": 0.8168,
+      "step": 415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1807580174927114e-05,
+      "loss": 0.6048,
+      "step": 416
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.176384839650146e-05,
+      "loss": 0.681,
+      "step": 417
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1720116618075802e-05,
+      "loss": 0.779,
+      "step": 418
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1676384839650146e-05,
+      "loss": 0.7463,
+      "step": 419
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.163265306122449e-05,
+      "loss": 0.7479,
+      "step": 420
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1588921282798833e-05,
+      "loss": 0.7277,
+      "step": 421
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1545189504373177e-05,
+      "loss": 0.7327,
+      "step": 422
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1501457725947523e-05,
+      "loss": 0.6682,
+      "step": 423
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1457725947521867e-05,
+      "loss": 0.6823,
+      "step": 424
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1413994169096211e-05,
+      "loss": 0.6142,
+      "step": 425
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1370262390670555e-05,
+      "loss": 0.7939,
+      "step": 426
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1326530612244897e-05,
+      "loss": 0.7539,
+      "step": 427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1282798833819241e-05,
+      "loss": 0.6073,
+      "step": 428
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1239067055393587e-05,
+      "loss": 0.7128,
+      "step": 429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.119533527696793e-05,
+      "loss": 0.6475,
+      "step": 430
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1151603498542275e-05,
+      "loss": 0.7107,
+      "step": 431
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1107871720116619e-05,
+      "loss": 0.6458,
+      "step": 432
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1064139941690962e-05,
+      "loss": 0.8056,
+      "step": 433
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.7321,
+      "step": 434
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.097667638483965e-05,
+      "loss": 0.7666,
+      "step": 435
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0932944606413994e-05,
+      "loss": 0.7894,
+      "step": 436
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0889212827988338e-05,
+      "loss": 0.581,
+      "step": 437
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0845481049562682e-05,
+      "loss": 0.6994,
+      "step": 438
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0801749271137026e-05,
+      "loss": 0.7405,
+      "step": 439
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.075801749271137e-05,
+      "loss": 0.7731,
+      "step": 440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.8333,
+      "step": 441
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.067055393586006e-05,
+      "loss": 0.8215,
+      "step": 442
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0626822157434404e-05,
+      "loss": 0.7639,
+      "step": 443
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0583090379008748e-05,
+      "loss": 0.6595,
+      "step": 444
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.053935860058309e-05,
+      "loss": 0.7345,
+      "step": 445
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0495626822157434e-05,
+      "loss": 0.6349,
+      "step": 446
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0451895043731778e-05,
+      "loss": 0.7739,
+      "step": 447
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0408163265306123e-05,
+      "loss": 0.7256,
+      "step": 448
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0364431486880467e-05,
+      "loss": 0.6909,
+      "step": 449
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0320699708454811e-05,
+      "loss": 0.663,
+      "step": 450
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0276967930029155e-05,
+      "loss": 0.6206,
+      "step": 451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.02332361516035e-05,
+      "loss": 0.6993,
+      "step": 452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0189504373177841e-05,
+      "loss": 0.7027,
+      "step": 453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0145772594752185e-05,
+      "loss": 0.7586,
+      "step": 454
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0102040816326531e-05,
+      "loss": 0.821,
+      "step": 455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0058309037900875e-05,
+      "loss": 0.7111,
+      "step": 456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0014577259475219e-05,
+      "loss": 0.6358,
+      "step": 457
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.970845481049563e-06,
+      "loss": 0.635,
+      "step": 458
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.927113702623907e-06,
+      "loss": 0.6775,
+      "step": 459
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.88338192419825e-06,
+      "loss": 0.6923,
+      "step": 460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.839650145772596e-06,
+      "loss": 0.7741,
+      "step": 461
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 0.6225,
+      "step": 462
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.752186588921283e-06,
+      "loss": 0.733,
+      "step": 463
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.708454810495627e-06,
+      "loss": 0.8117,
+      "step": 464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.66472303206997e-06,
+      "loss": 0.7334,
+      "step": 465
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.620991253644314e-06,
+      "loss": 0.6461,
+      "step": 466
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.57725947521866e-06,
+      "loss": 0.7507,
+      "step": 467
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.533527696793004e-06,
+      "loss": 0.6619,
+      "step": 468
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.489795918367348e-06,
+      "loss": 0.6141,
+      "step": 469
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.446064139941692e-06,
+      "loss": 0.6757,
+      "step": 470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.402332361516034e-06,
+      "loss": 0.6319,
+      "step": 471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.358600583090378e-06,
+      "loss": 0.651,
+      "step": 472
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.314868804664722e-06,
+      "loss": 0.6266,
+      "step": 473
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.271137026239068e-06,
+      "loss": 0.5657,
+      "step": 474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.227405247813412e-06,
+      "loss": 0.6838,
+      "step": 475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 0.8081,
+      "step": 476
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.1399416909621e-06,
+      "loss": 0.5915,
+      "step": 477
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.096209912536444e-06,
+      "loss": 0.6921,
+      "step": 478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.052478134110787e-06,
+      "loss": 0.6148,
+      "step": 479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.008746355685131e-06,
+      "loss": 0.6602,
+      "step": 480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.965014577259475e-06,
+      "loss": 0.5945,
+      "step": 481
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.92128279883382e-06,
+      "loss": 0.6965,
+      "step": 482
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.877551020408163e-06,
+      "loss": 0.7285,
+      "step": 483
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.833819241982507e-06,
+      "loss": 0.8048,
+      "step": 484
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.790087463556851e-06,
+      "loss": 0.6653,
+      "step": 485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.746355685131197e-06,
+      "loss": 0.6806,
+      "step": 486
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.70262390670554e-06,
+      "loss": 0.7117,
+      "step": 487
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.658892128279885e-06,
+      "loss": 0.6386,
+      "step": 488
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.615160349854227e-06,
+      "loss": 0.6763,
+      "step": 489
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.7101,
+      "step": 490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.527696793002915e-06,
+      "loss": 0.6231,
+      "step": 491
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.483965014577259e-06,
+      "loss": 0.6945,
+      "step": 492
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.440233236151604e-06,
+      "loss": 0.6969,
+      "step": 493
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.396501457725948e-06,
+      "loss": 0.6003,
+      "step": 494
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.352769679300292e-06,
+      "loss": 0.7275,
+      "step": 495
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.309037900874636e-06,
+      "loss": 0.659,
+      "step": 496
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.265306122448979e-06,
+      "loss": 0.6436,
+      "step": 497
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.221574344023323e-06,
+      "loss": 0.7147,
+      "step": 498
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.177842565597668e-06,
+      "loss": 0.6923,
+      "step": 499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.134110787172012e-06,
+      "loss": 0.6701,
+      "step": 500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.090379008746356e-06,
+      "loss": 0.7177,
+      "step": 501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.0466472303207e-06,
+      "loss": 0.6963,
+      "step": 502
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.002915451895044e-06,
+      "loss": 0.7587,
+      "step": 503
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.959183673469388e-06,
+      "loss": 0.5824,
+      "step": 504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.915451895043732e-06,
+      "loss": 0.7099,
+      "step": 505
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.871720116618076e-06,
+      "loss": 0.6534,
+      "step": 506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.82798833819242e-06,
+      "loss": 0.6332,
+      "step": 507
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.784256559766764e-06,
+      "loss": 0.655,
+      "step": 508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.740524781341108e-06,
+      "loss": 0.6822,
+      "step": 509
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.696793002915452e-06,
+      "loss": 0.5835,
+      "step": 510
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.653061224489796e-06,
+      "loss": 0.7233,
+      "step": 511
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.60932944606414e-06,
+      "loss": 0.6536,
+      "step": 512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.565597667638485e-06,
+      "loss": 0.6018,
+      "step": 513
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.521865889212829e-06,
+      "loss": 0.6019,
+      "step": 514
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.478134110787172e-06,
+      "loss": 0.7141,
+      "step": 515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.434402332361516e-06,
+      "loss": 0.7671,
+      "step": 516
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.39067055393586e-06,
+      "loss": 0.6633,
+      "step": 517
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.346938775510204e-06,
+      "loss": 0.5977,
+      "step": 518
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.303206997084548e-06,
+      "loss": 0.7577,
+      "step": 519
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.259475218658893e-06,
+      "loss": 0.7126,
+      "step": 520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.215743440233237e-06,
+      "loss": 0.6572,
+      "step": 521
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.17201166180758e-06,
+      "loss": 0.6463,
+      "step": 522
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.128279883381925e-06,
+      "loss": 0.565,
+      "step": 523
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.0845481049562685e-06,
+      "loss": 0.6441,
+      "step": 524
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.0408163265306125e-06,
+      "loss": 0.8821,
+      "step": 525
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.9970845481049564e-06,
+      "loss": 0.6033,
+      "step": 526
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.9533527696793e-06,
+      "loss": 0.6548,
+      "step": 527
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.909620991253644e-06,
+      "loss": 0.6689,
+      "step": 528
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.865889212827989e-06,
+      "loss": 0.6853,
+      "step": 529
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.822157434402332e-06,
+      "loss": 0.5613,
+      "step": 530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.778425655976676e-06,
+      "loss": 0.7409,
+      "step": 531
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.734693877551021e-06,
+      "loss": 0.644,
+      "step": 532
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.690962099125365e-06,
+      "loss": 0.5878,
+      "step": 533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.647230320699709e-06,
+      "loss": 0.6137,
+      "step": 534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.603498542274052e-06,
+      "loss": 0.7217,
+      "step": 535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.559766763848397e-06,
+      "loss": 0.7005,
+      "step": 536
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.516034985422741e-06,
+      "loss": 0.7718,
+      "step": 537
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.472303206997085e-06,
+      "loss": 0.6834,
+      "step": 538
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.428571428571429e-06,
+      "loss": 0.6291,
+      "step": 539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.3848396501457726e-06,
+      "loss": 0.6906,
+      "step": 540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.3411078717201165e-06,
+      "loss": 0.7809,
+      "step": 541
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.297376093294461e-06,
+      "loss": 0.661,
+      "step": 542
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.253644314868805e-06,
+      "loss": 0.6932,
+      "step": 543
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.209912536443148e-06,
+      "loss": 0.7602,
+      "step": 544
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.166180758017493e-06,
+      "loss": 0.7317,
+      "step": 545
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 0.7391,
+      "step": 546
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.078717201166181e-06,
+      "loss": 0.6777,
+      "step": 547
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.034985422740525e-06,
+      "loss": 0.7403,
+      "step": 548
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.991253644314869e-06,
+      "loss": 0.6889,
+      "step": 549
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.947521865889213e-06,
+      "loss": 0.655,
+      "step": 550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.903790087463557e-06,
+      "loss": 0.5358,
+      "step": 551
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.860058309037901e-06,
+      "loss": 0.6668,
+      "step": 552
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.816326530612245e-06,
+      "loss": 0.5347,
+      "step": 553
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.772594752186589e-06,
+      "loss": 0.5213,
+      "step": 554
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.7288629737609335e-06,
+      "loss": 0.7983,
+      "step": 555
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6851311953352774e-06,
+      "loss": 0.7329,
+      "step": 556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6413994169096205e-06,
+      "loss": 0.6845,
+      "step": 557
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.597667638483965e-06,
+      "loss": 0.54,
+      "step": 558
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.553935860058309e-06,
+      "loss": 0.5828,
+      "step": 559
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.510204081632653e-06,
+      "loss": 0.7079,
+      "step": 560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.466472303206997e-06,
+      "loss": 0.6471,
+      "step": 561
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.422740524781341e-06,
+      "loss": 0.7349,
+      "step": 562
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.379008746355685e-06,
+      "loss": 0.6399,
+      "step": 563
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.33527696793003e-06,
+      "loss": 0.7541,
+      "step": 564
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.291545189504374e-06,
+      "loss": 0.6564,
+      "step": 565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.247813411078717e-06,
+      "loss": 0.5897,
+      "step": 566
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.204081632653062e-06,
+      "loss": 0.7713,
+      "step": 567
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.160349854227406e-06,
+      "loss": 0.5714,
+      "step": 568
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.11661807580175e-06,
+      "loss": 0.666,
+      "step": 569
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.072886297376093e-06,
+      "loss": 0.6647,
+      "step": 570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0291545189504375e-06,
+      "loss": 0.7072,
+      "step": 571
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9854227405247814e-06,
+      "loss": 0.7326,
+      "step": 572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.941690962099125e-06,
+      "loss": 0.6275,
+      "step": 573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 0.6175,
+      "step": 574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.854227405247813e-06,
+      "loss": 0.6715,
+      "step": 575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.810495626822157e-06,
+      "loss": 0.69,
+      "step": 576
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.766763848396502e-06,
+      "loss": 0.6194,
+      "step": 577
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.723032069970846e-06,
+      "loss": 0.7825,
+      "step": 578
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.679300291545189e-06,
+      "loss": 0.5966,
+      "step": 579
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.635568513119534e-06,
+      "loss": 0.593,
+      "step": 580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 0.8133,
+      "step": 581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.548104956268222e-06,
+      "loss": 0.665,
+      "step": 582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.504373177842566e-06,
+      "loss": 0.6998,
+      "step": 583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.46064139941691e-06,
+      "loss": 0.6609,
+      "step": 584
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.416909620991254e-06,
+      "loss": 0.5941,
+      "step": 585
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.373177842565598e-06,
+      "loss": 0.6983,
+      "step": 586
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.329446064139942e-06,
+      "loss": 0.5953,
+      "step": 587
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.7267,
+      "step": 588
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.241982507288629e-06,
+      "loss": 0.791,
+      "step": 589
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.198250728862974e-06,
+      "loss": 0.6921,
+      "step": 590
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.154518950437318e-06,
+      "loss": 0.8183,
+      "step": 591
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.110787172011661e-06,
+      "loss": 0.6162,
+      "step": 592
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.067055393586006e-06,
+      "loss": 0.7695,
+      "step": 593
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.02332361516035e-06,
+      "loss": 0.6925,
+      "step": 594
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.979591836734694e-06,
+      "loss": 0.6336,
+      "step": 595
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.935860058309038e-06,
+      "loss": 0.7375,
+      "step": 596
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.892128279883382e-06,
+      "loss": 0.8469,
+      "step": 597
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.848396501457726e-06,
+      "loss": 0.635,
+      "step": 598
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.80466472303207e-06,
+      "loss": 0.9703,
+      "step": 599
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7609329446064145e-06,
+      "loss": 0.653,
+      "step": 600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.717201166180758e-06,
+      "loss": 0.6354,
+      "step": 601
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.673469387755102e-06,
+      "loss": 0.6025,
+      "step": 602
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6297376093294464e-06,
+      "loss": 0.6416,
+      "step": 603
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.58600583090379e-06,
+      "loss": 0.6034,
+      "step": 604
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5422740524781343e-06,
+      "loss": 0.6376,
+      "step": 605
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4985422740524782e-06,
+      "loss": 0.6776,
+      "step": 606
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.454810495626822e-06,
+      "loss": 0.6841,
+      "step": 607
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.411078717201166e-06,
+      "loss": 0.641,
+      "step": 608
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3673469387755105e-06,
+      "loss": 0.76,
+      "step": 609
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3236151603498544e-06,
+      "loss": 0.7558,
+      "step": 610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2798833819241984e-06,
+      "loss": 0.6344,
+      "step": 611
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2361516034985423e-06,
+      "loss": 0.6477,
+      "step": 612
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1924198250728863e-06,
+      "loss": 0.5704,
+      "step": 613
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1486880466472307e-06,
+      "loss": 0.7345,
+      "step": 614
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.104956268221574e-06,
+      "loss": 0.6538,
+      "step": 615
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 0.6457,
+      "step": 616
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0174927113702625e-06,
+      "loss": 0.6632,
+      "step": 617
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9737609329446064e-06,
+      "loss": 0.5877,
+      "step": 618
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9300291545189504e-06,
+      "loss": 0.7263,
+      "step": 619
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8862973760932943e-06,
+      "loss": 0.7151,
+      "step": 620
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8425655976676387e-06,
+      "loss": 0.7417,
+      "step": 621
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7988338192419827e-06,
+      "loss": 0.7464,
+      "step": 622
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7551020408163266e-06,
+      "loss": 0.6743,
+      "step": 623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7113702623906706e-06,
+      "loss": 0.6379,
+      "step": 624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.667638483965015e-06,
+      "loss": 0.6686,
+      "step": 625
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6239067055393585e-06,
+      "loss": 0.6287,
+      "step": 626
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.580174927113703e-06,
+      "loss": 0.8058,
+      "step": 627
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5364431486880463e-06,
+      "loss": 0.6915,
+      "step": 628
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4927113702623907e-06,
+      "loss": 0.6038,
+      "step": 629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 0.6473,
+      "step": 630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4052478134110786e-06,
+      "loss": 0.6237,
+      "step": 631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.361516034985423e-06,
+      "loss": 0.8068,
+      "step": 632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.317784256559767e-06,
+      "loss": 0.6403,
+      "step": 633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.274052478134111e-06,
+      "loss": 0.6132,
+      "step": 634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.230320699708455e-06,
+      "loss": 0.8262,
+      "step": 635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.186588921282799e-06,
+      "loss": 0.7788,
+      "step": 636
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.6717,
+      "step": 637
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.099125364431487e-06,
+      "loss": 0.7498,
+      "step": 638
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0553935860058306e-06,
+      "loss": 0.6071,
+      "step": 639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.011661807580175e-06,
+      "loss": 0.6848,
+      "step": 640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.967930029154519e-06,
+      "loss": 0.784,
+      "step": 641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.924198250728863e-06,
+      "loss": 0.6056,
+      "step": 642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8804664723032073e-06,
+      "loss": 0.7672,
+      "step": 643
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.836734693877551e-06,
+      "loss": 0.7579,
+      "step": 644
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.793002915451895e-06,
+      "loss": 0.6712,
+      "step": 645
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7492711370262391e-06,
+      "loss": 0.758,
+      "step": 646
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.705539358600583e-06,
+      "loss": 0.6262,
+      "step": 647
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6618075801749272e-06,
+      "loss": 0.5719,
+      "step": 648
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6180758017492712e-06,
+      "loss": 0.6228,
+      "step": 649
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5743440233236153e-06,
+      "loss": 0.5954,
+      "step": 650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5306122448979593e-06,
+      "loss": 0.6157,
+      "step": 651
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4868804664723032e-06,
+      "loss": 0.6907,
+      "step": 652
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4431486880466472e-06,
+      "loss": 0.5488,
+      "step": 653
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3994169096209913e-06,
+      "loss": 0.5904,
+      "step": 654
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3556851311953353e-06,
+      "loss": 0.608,
+      "step": 655
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3119533527696792e-06,
+      "loss": 0.8896,
+      "step": 656
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2682215743440232e-06,
+      "loss": 0.6197,
+      "step": 657
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 0.6063,
+      "step": 658
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1807580174927115e-06,
+      "loss": 0.7669,
+      "step": 659
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1370262390670554e-06,
+      "loss": 0.8527,
+      "step": 660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0932944606413996e-06,
+      "loss": 0.6512,
+      "step": 661
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0495626822157436e-06,
+      "loss": 0.6591,
+      "step": 662
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0058309037900875e-06,
+      "loss": 0.6701,
+      "step": 663
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.620991253644314e-07,
+      "loss": 0.598,
+      "step": 664
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.183673469387755e-07,
+      "loss": 0.7497,
+      "step": 665
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.746355685131196e-07,
+      "loss": 0.6544,
+      "step": 666
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.309037900874636e-07,
+      "loss": 0.7283,
+      "step": 667
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.871720116618077e-07,
+      "loss": 0.7193,
+      "step": 668
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.434402332361516e-07,
+      "loss": 0.6521,
+      "step": 669
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.997084548104957e-07,
+      "loss": 0.5521,
+      "step": 670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.559766763848396e-07,
+      "loss": 0.581,
+      "step": 671
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 0.6203,
+      "step": 672
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.685131195335277e-07,
+      "loss": 0.5413,
+      "step": 673
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.247813411078718e-07,
+      "loss": 0.7687,
+      "step": 674
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.810495626822157e-07,
+      "loss": 0.6807,
+      "step": 675
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.373177842565598e-07,
+      "loss": 0.6888,
+      "step": 676
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.9358600583090383e-07,
+      "loss": 0.59,
+      "step": 677
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4985422740524783e-07,
+      "loss": 0.617,
+      "step": 678
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0612244897959183e-07,
+      "loss": 0.6244,
+      "step": 679
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.623906705539359e-07,
+      "loss": 0.6787,
+      "step": 680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.186588921282799e-07,
+      "loss": 0.7535,
+      "step": 681
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7492711370262392e-07,
+      "loss": 0.649,
+      "step": 682
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3119533527696794e-07,
+      "loss": 0.6996,
+      "step": 683
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.746355685131196e-08,
+      "loss": 0.7669,
+      "step": 684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.373177842565598e-08,
+      "loss": 0.7109,
+      "step": 685
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.7227,
+      "step": 686
+    },
+    {
+      "epoch": 2.0,
+      "step": 686,
+      "total_flos": 1.932278221504512e+17,
+      "train_loss": 0.8650754356088861,
+      "train_runtime": 3796.8514,
+      "train_samples_per_second": 46.143,
+      "train_steps_per_second": 0.181
+    }
+  ],
+  "max_steps": 686,
+  "num_train_epochs": 2,
+  "total_flos": 1.932278221504512e+17,
+  "trial_name": null,
+  "trial_params": null
+}