diff --git "a/lora-checkpoints-7/trainer_state.json" "b/lora-checkpoints-7/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/lora-checkpoints-7/trainer_state.json"
@@ -0,0 +1,52069 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 8674,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.662835249042146e-08,
+      "loss": 1.1546,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5325670498084292e-07,
+      "loss": 1.5863,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2988505747126437e-07,
+      "loss": 1.7081,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0651340996168583e-07,
+      "loss": 1.1316,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.831417624521073e-07,
+      "loss": 1.1345,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5977011494252875e-07,
+      "loss": 1.3013,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.363984674329502e-07,
+      "loss": 1.2092,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.130268199233717e-07,
+      "loss": 1.1024,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.896551724137931e-07,
+      "loss": 1.3497,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.662835249042146e-07,
+      "loss": 1.3732,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.429118773946361e-07,
+      "loss": 1.258,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.195402298850575e-07,
+      "loss": 1.2873,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.96168582375479e-07,
+      "loss": 1.1228,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0727969348659004e-06,
+      "loss": 1.688,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1494252873563219e-06,
+      "loss": 1.1013,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2260536398467433e-06,
+      "loss": 1.1602,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3026819923371648e-06,
+      "loss": 1.6803,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 1.2273,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.455938697318008e-06,
+      "loss": 1.1842,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5325670498084292e-06,
+      "loss": 1.1551,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6091954022988506e-06,
+      "loss": 1.1242,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6858237547892723e-06,
+      "loss": 1.1441,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7624521072796935e-06,
+      "loss": 1.1528,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.839080459770115e-06,
+      "loss": 1.1718,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9157088122605367e-06,
+      "loss": 1.5361,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.992337164750958e-06,
+      "loss": 0.8994,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0689655172413796e-06,
+      "loss": 1.7273,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.145593869731801e-06,
+      "loss": 1.4909,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.073,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2988505747126437e-06,
+      "loss": 1.0684,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3754789272030654e-06,
+      "loss": 1.1498,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4521072796934867e-06,
+      "loss": 1.1492,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5287356321839083e-06,
+      "loss": 1.8386,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6053639846743296e-06,
+      "loss": 1.1221,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6819923371647512e-06,
+      "loss": 1.2481,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7586206896551725e-06,
+      "loss": 1.3012,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.835249042145594e-06,
+      "loss": 1.2322,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.911877394636016e-06,
+      "loss": 1.1269,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.988505747126437e-06,
+      "loss": 1.241,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0651340996168583e-06,
+      "loss": 1.2431,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.14176245210728e-06,
+      "loss": 1.3121,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2183908045977012e-06,
+      "loss": 1.2629,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.295019157088123e-06,
+      "loss": 1.1583,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3716475095785446e-06,
+      "loss": 1.2415,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 1.0578,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.524904214559387e-06,
+      "loss": 1.4564,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6015325670498087e-06,
+      "loss": 1.7213,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.67816091954023e-06,
+      "loss": 1.037,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7547892720306517e-06,
+      "loss": 1.0886,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.831417624521073e-06,
+      "loss": 1.4683,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.908045977011495e-06,
+      "loss": 1.2022,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.984674329501916e-06,
+      "loss": 0.9431,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0613026819923375e-06,
+      "loss": 1.0793,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.137931034482759e-06,
+      "loss": 1.144,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.214559386973181e-06,
+      "loss": 1.1128,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.291187739463602e-06,
+      "loss": 1.0449,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.367816091954023e-06,
+      "loss": 1.1486,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.0338,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.521072796934866e-06,
+      "loss": 1.1624,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5977011494252875e-06,
+      "loss": 1.0294,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.674329501915709e-06,
+      "loss": 1.0004,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.750957854406131e-06,
+      "loss": 1.2661,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8275862068965525e-06,
+      "loss": 0.8371,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.904214559386973e-06,
+      "loss": 0.9805,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980842911877395e-06,
+      "loss": 0.9499,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.057471264367817e-06,
+      "loss": 1.104,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.134099616858238e-06,
+      "loss": 1.0426,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.210727969348659e-06,
+      "loss": 1.0808,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.287356321839081e-06,
+      "loss": 1.0336,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3639846743295025e-06,
+      "loss": 1.1224,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.440613026819924e-06,
+      "loss": 0.9784,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.517241379310345e-06,
+      "loss": 1.5087,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.593869731800766e-06,
+      "loss": 1.1665,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.670498084291188e-06,
+      "loss": 1.4407,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.747126436781609e-06,
+      "loss": 1.1289,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.823754789272032e-06,
+      "loss": 1.078,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9003831417624525e-06,
+      "loss": 1.4028,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.977011494252874e-06,
+      "loss": 0.954,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.053639846743296e-06,
+      "loss": 0.9342,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.130268199233717e-06,
+      "loss": 1.1133,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.206896551724138e-06,
+      "loss": 1.0641,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.28352490421456e-06,
+      "loss": 1.069,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.360153256704982e-06,
+      "loss": 1.0133,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4367816091954025e-06,
+      "loss": 1.2944,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.513409961685824e-06,
+      "loss": 0.9581,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.590038314176246e-06,
+      "loss": 1.0384,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0838,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.743295019157089e-06,
+      "loss": 0.9878,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.81992337164751e-06,
+      "loss": 0.9764,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 0.9388,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.973180076628353e-06,
+      "loss": 1.0298,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.049808429118774e-06,
+      "loss": 0.8262,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.126436781609196e-06,
+      "loss": 0.9274,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2030651340996175e-06,
+      "loss": 1.0444,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.279693486590039e-06,
+      "loss": 0.9313,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.35632183908046e-06,
+      "loss": 1.0876,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4329501915708825e-06,
+      "loss": 0.8926,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.509578544061303e-06,
+      "loss": 1.1281,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.586206896551724e-06,
+      "loss": 0.9393,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.662835249042147e-06,
+      "loss": 0.9635,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.739463601532567e-06,
+      "loss": 0.8897,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.81609195402299e-06,
+      "loss": 0.9278,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.89272030651341e-06,
+      "loss": 0.8852,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.969348659003832e-06,
+      "loss": 0.9289,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.045977011494253e-06,
+      "loss": 0.8855,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.122605363984675e-06,
+      "loss": 0.8429,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.199233716475097e-06,
+      "loss": 1.0174,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.275862068965518e-06,
+      "loss": 1.1034,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.35249042145594e-06,
+      "loss": 0.9461,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.429118773946362e-06,
+      "loss": 0.9861,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.505747126436782e-06,
+      "loss": 0.7808,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.582375478927203e-06,
+      "loss": 1.0249,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.659003831417625e-06,
+      "loss": 1.2094,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.735632183908047e-06,
+      "loss": 0.8701,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.812260536398468e-06,
+      "loss": 0.8878,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.1241,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.965517241379312e-06,
+      "loss": 0.9902,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.042145593869732e-06,
+      "loss": 0.9316,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.118773946360155e-06,
+      "loss": 1.3103,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.195402298850575e-06,
+      "loss": 0.873,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.272030651340997e-06,
+      "loss": 1.0002,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.348659003831418e-06,
+      "loss": 1.243,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.42528735632184e-06,
+      "loss": 0.9723,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.501915708812262e-06,
+      "loss": 0.8332,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.578544061302683e-06,
+      "loss": 1.0264,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.655172413793105e-06,
+      "loss": 0.8781,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.731800766283525e-06,
+      "loss": 1.1966,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.808429118773947e-06,
+      "loss": 0.8858,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.885057471264368e-06,
+      "loss": 1.0825,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.96168582375479e-06,
+      "loss": 0.8783,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0038314176245212e-05,
+      "loss": 1.0097,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0114942528735633e-05,
+      "loss": 1.198,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0191570881226053e-05,
+      "loss": 1.0063,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0268199233716477e-05,
+      "loss": 1.0284,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0344827586206898e-05,
+      "loss": 0.9751,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0421455938697318e-05,
+      "loss": 1.423,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.049808429118774e-05,
+      "loss": 0.9304,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0574712643678162e-05,
+      "loss": 0.8565,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0651340996168585e-05,
+      "loss": 1.2994,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0727969348659005e-05,
+      "loss": 0.8531,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0804597701149427e-05,
+      "loss": 0.882,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0881226053639848e-05,
+      "loss": 1.1846,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0957854406130268e-05,
+      "loss": 0.768,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.103448275862069e-05,
+      "loss": 0.9345,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.2173,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1187739463601532e-05,
+      "loss": 0.9213,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1264367816091955e-05,
+      "loss": 0.9143,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1340996168582377e-05,
+      "loss": 0.8126,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1417624521072798e-05,
+      "loss": 0.807,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1494252873563218e-05,
+      "loss": 0.9815,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.157088122605364e-05,
+      "loss": 0.7839,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1647509578544063e-05,
+      "loss": 0.823,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1724137931034483e-05,
+      "loss": 0.9838,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1800766283524905e-05,
+      "loss": 0.8245,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1877394636015327e-05,
+      "loss": 0.8341,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1954022988505748e-05,
+      "loss": 0.8138,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2030651340996168e-05,
+      "loss": 0.9946,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2107279693486592e-05,
+      "loss": 0.8,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2183908045977013e-05,
+      "loss": 0.9336,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2260536398467433e-05,
+      "loss": 0.7798,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2337164750957855e-05,
+      "loss": 0.8441,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2413793103448277e-05,
+      "loss": 0.8736,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.24904214559387e-05,
+      "loss": 0.9802,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.256704980842912e-05,
+      "loss": 0.8886,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2643678160919542e-05,
+      "loss": 0.8765,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2720306513409963e-05,
+      "loss": 0.8983,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2796934865900383e-05,
+      "loss": 1.1707,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2873563218390805e-05,
+      "loss": 0.9825,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2950191570881228e-05,
+      "loss": 1.2334,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3026819923371648e-05,
+      "loss": 0.9916,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.310344827586207e-05,
+      "loss": 1.0685,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3180076628352492e-05,
+      "loss": 1.1044,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3256704980842913e-05,
+      "loss": 1.2189,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1223,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3409961685823755e-05,
+      "loss": 0.7837,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3486590038314178e-05,
+      "loss": 0.8925,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3563218390804598e-05,
+      "loss": 0.7335,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.363984674329502e-05,
+      "loss": 0.9763,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3716475095785442e-05,
+      "loss": 0.8523,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 0.8323,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3869731800766283e-05,
+      "loss": 0.9419,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3946360153256707e-05,
+      "loss": 0.825,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4022988505747128e-05,
+      "loss": 0.9965,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4099616858237548e-05,
+      "loss": 1.1357,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.417624521072797e-05,
+      "loss": 0.753,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4252873563218392e-05,
+      "loss": 0.8237,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4329501915708815e-05,
+      "loss": 0.7998,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4406130268199235e-05,
+      "loss": 0.7095,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4482758620689657e-05,
+      "loss": 0.8841,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4559386973180078e-05,
+      "loss": 0.9466,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4636015325670498e-05,
+      "loss": 0.9332,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.471264367816092e-05,
+      "loss": 0.881,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4789272030651343e-05,
+      "loss": 1.1389,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4865900383141765e-05,
+      "loss": 1.0423,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4942528735632185e-05,
+      "loss": 0.8445,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5019157088122607e-05,
+      "loss": 0.757,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5095785440613028e-05,
+      "loss": 0.7793,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5172413793103448e-05,
+      "loss": 0.8129,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5249042145593872e-05,
+      "loss": 0.8049,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5325670498084293e-05,
+      "loss": 0.8567,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.540229885057471e-05,
+      "loss": 1.2996,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5478927203065133e-05,
+      "loss": 0.761,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.7086,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.563218390804598e-05,
+      "loss": 0.8677,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5708812260536398e-05,
+      "loss": 0.8787,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.578544061302682e-05,
+      "loss": 0.8306,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.586206896551724e-05,
+      "loss": 0.7043,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5938697318007663e-05,
+      "loss": 0.9626,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6015325670498085e-05,
+      "loss": 1.2582,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6091954022988507e-05,
+      "loss": 0.8121,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6168582375478928e-05,
+      "loss": 0.869,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.624521072796935e-05,
+      "loss": 0.935,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.632183908045977e-05,
+      "loss": 0.7647,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6398467432950193e-05,
+      "loss": 0.813,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6475095785440615e-05,
+      "loss": 0.7999,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6551724137931037e-05,
+      "loss": 0.8408,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.662835249042146e-05,
+      "loss": 0.7241,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.670498084291188e-05,
+      "loss": 1.0742,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6781609195402298e-05,
+      "loss": 0.6646,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6858237547892723e-05,
+      "loss": 0.8214,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6934865900383145e-05,
+      "loss": 1.098,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7011494252873563e-05,
+      "loss": 1.2818,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7088122605363985e-05,
+      "loss": 0.7909,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7164750957854407e-05,
+      "loss": 1.2357,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 1.1155,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.731800766283525e-05,
+      "loss": 1.0863,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.739463601532567e-05,
+      "loss": 1.2157,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7471264367816093e-05,
+      "loss": 0.8436,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7547892720306515e-05,
+      "loss": 0.8008,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7624521072796937e-05,
+      "loss": 1.235,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.770114942528736e-05,
+      "loss": 0.7781,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.8425,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.78544061302682e-05,
+      "loss": 0.8809,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7931034482758623e-05,
+      "loss": 1.2836,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8007662835249045e-05,
+      "loss": 0.8128,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8084291187739463e-05,
+      "loss": 0.8046,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8160919540229885e-05,
+      "loss": 0.8212,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.823754789272031e-05,
+      "loss": 0.9662,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8314176245210728e-05,
+      "loss": 0.9169,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.839080459770115e-05,
+      "loss": 0.9057,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.846743295019157e-05,
+      "loss": 0.8638,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8544061302681993e-05,
+      "loss": 0.8866,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8620689655172415e-05,
+      "loss": 0.8727,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8697318007662837e-05,
+      "loss": 0.8866,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8773946360153258e-05,
+      "loss": 0.9735,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.885057471264368e-05,
+      "loss": 1.2856,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.89272030651341e-05,
+      "loss": 0.9178,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9003831417624523e-05,
+      "loss": 0.825,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.908045977011494e-05,
+      "loss": 1.321,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9157088122605367e-05,
+      "loss": 1.2795,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.923371647509579e-05,
+      "loss": 0.8104,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.931034482758621e-05,
+      "loss": 1.2033,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9386973180076628e-05,
+      "loss": 0.7198,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.946360153256705e-05,
+      "loss": 0.8344,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9540229885057475e-05,
+      "loss": 0.7676,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9616858237547893e-05,
+      "loss": 0.903,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9693486590038315e-05,
+      "loss": 1.1091,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9770114942528737e-05,
+      "loss": 0.9962,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846743295019158e-05,
+      "loss": 0.7868,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.992337164750958e-05,
+      "loss": 0.8588,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.7914,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999999302783e-05,
+      "loss": 0.8769,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999997211132103e-05,
+      "loss": 0.6922,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999993725047598e-05,
+      "loss": 0.926,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999998884452997e-05,
+      "loss": 0.8409,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999982569579905e-05,
+      "loss": 1.3052,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999997490019827e-05,
+      "loss": 1.1054,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999965836386135e-05,
+      "loss": 0.7954,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999995537814477e-05,
+      "loss": 0.897,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999943525475628e-05,
+      "loss": 0.8755,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999930278380367e-05,
+      "loss": 0.7123,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999991563686083e-05,
+      "loss": 1.1409,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999989960091906e-05,
+      "loss": 1.1365,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999882170557293e-05,
+      "loss": 0.7635,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999863345777962e-05,
+      "loss": 1.0736,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999843126583686e-05,
+      "loss": 0.8603,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999982151297729e-05,
+      "loss": 0.8732,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999798504961787e-05,
+      "loss": 0.8123,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999774102540382e-05,
+      "loss": 0.768,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999974830571648e-05,
+      "loss": 0.9274,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999972111449368e-05,
+      "loss": 0.7464,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999969252887577e-05,
+      "loss": 0.9448,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999662548866742e-05,
+      "loss": 0.7869,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999631174470768e-05,
+      "loss": 0.927,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999959840569223e-05,
+      "loss": 1.0515,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999564242535693e-05,
+      "loss": 0.7353,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999528685005926e-05,
+      "loss": 0.7149,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999949173310788e-05,
+      "loss": 1.141,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999945338684671e-05,
+      "loss": 0.7104,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999413646227767e-05,
+      "loss": 1.0244,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999372511256592e-05,
+      "loss": 0.8683,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999932998193892e-05,
+      "loss": 0.8275,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999286058280676e-05,
+      "loss": 0.8709,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999240740287994e-05,
+      "loss": 0.7817,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999194027967188e-05,
+      "loss": 1.0609,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999145921324773e-05,
+      "loss": 0.8709,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999096420367455e-05,
+      "loss": 0.8528,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999045525102137e-05,
+      "loss": 1.0058,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998993235535922e-05,
+      "loss": 0.9185,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999893955167609e-05,
+      "loss": 0.6547,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998884473530143e-05,
+      "loss": 0.8247,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998828001105743e-05,
+      "loss": 0.7599,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999877013441078e-05,
+      "loss": 0.8533,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998710873453314e-05,
+      "loss": 0.9897,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998650218241612e-05,
+      "loss": 1.2307,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999858816878413e-05,
+      "loss": 0.8441,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998524725089524e-05,
+      "loss": 0.8947,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998459887166635e-05,
+      "loss": 0.8329,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999839365502451e-05,
+      "loss": 1.0646,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998326028672383e-05,
+      "loss": 1.1742,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998257008119682e-05,
+      "loss": 1.2854,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998186593376032e-05,
+      "loss": 0.7544,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998114784451252e-05,
+      "loss": 0.7267,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999804158135536e-05,
+      "loss": 0.7987,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999796698409856e-05,
+      "loss": 1.1021,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999789099269125e-05,
+      "loss": 1.1815,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999781360714403e-05,
+      "loss": 0.7316,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997734827467692e-05,
+      "loss": 0.823,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997654653673222e-05,
+      "loss": 0.9557,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99975730857718e-05,
+      "loss": 1.0687,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997490123774792e-05,
+      "loss": 1.2387,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999740576769378e-05,
+      "loss": 0.7668,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999732001754051e-05,
+      "loss": 0.8425,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999723287332696e-05,
+      "loss": 1.0088,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997144335065264e-05,
+      "loss": 0.88,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997054402767774e-05,
+      "loss": 0.8113,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996963076447035e-05,
+      "loss": 0.7774,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996870356115775e-05,
+      "loss": 0.8445,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999677624178693e-05,
+      "loss": 0.7711,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996680733473618e-05,
+      "loss": 0.8834,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999658383118916e-05,
+      "loss": 0.7598,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996485534947066e-05,
+      "loss": 0.7547,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996385844761045e-05,
+      "loss": 0.7371,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996284760644995e-05,
+      "loss": 0.7496,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996182282613015e-05,
+      "loss": 0.8942,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996078410679398e-05,
+      "loss": 0.7809,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995973144858615e-05,
+      "loss": 0.8655,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999586648516536e-05,
+      "loss": 1.2124,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99957584316145e-05,
+      "loss": 1.1166,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995648984221097e-05,
+      "loss": 0.9669,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995538143000418e-05,
+      "loss": 0.7393,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995425907967925e-05,
+      "loss": 1.0752,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995312279139255e-05,
+      "loss": 0.7801,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995197256530267e-05,
+      "loss": 1.1855,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995080840156988e-05,
+      "loss": 0.6878,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994963030035656e-05,
+      "loss": 1.0031,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99948438261827e-05,
+      "loss": 0.7001,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999472322861474e-05,
+      "loss": 0.7116,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99946012373486e-05,
+      "loss": 1.1553,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994477852401276e-05,
+      "loss": 0.754,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994353073789988e-05,
+      "loss": 0.9099,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999422690153213e-05,
+      "loss": 1.1578,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994099335645295e-05,
+      "loss": 0.7803,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999397037614727e-05,
+      "loss": 1.1064,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993840023056045e-05,
+      "loss": 0.9806,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993708276389785e-05,
+      "loss": 0.8171,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993575136166868e-05,
+      "loss": 0.7291,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999344060240586e-05,
+      "loss": 0.755,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999330467512552e-05,
+      "loss": 0.717,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993167354344805e-05,
+      "loss": 0.9746,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993028640082855e-05,
+      "loss": 0.877,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999288853235902e-05,
+      "loss": 0.7695,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999274703119284e-05,
+      "loss": 1.0115,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992604136604036e-05,
+      "loss": 0.8283,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999245984861254e-05,
+      "loss": 1.1489,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992314167238476e-05,
+      "loss": 0.6715,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999216709250215e-05,
+      "loss": 0.7789,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992018624424073e-05,
+      "loss": 0.8714,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991868763024955e-05,
+      "loss": 0.8979,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999171750832568e-05,
+      "loss": 0.758,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999156486034735e-05,
+      "loss": 1.2209,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999141081911125e-05,
+      "loss": 0.7215,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991255384638857e-05,
+      "loss": 1.2372,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991098556951844e-05,
+      "loss": 0.7713,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990940336072083e-05,
+      "loss": 0.6957,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990780722021635e-05,
+      "loss": 0.7165,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999061971482276e-05,
+      "loss": 0.8867,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990457314497906e-05,
+      "loss": 0.6823,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990293521069714e-05,
+      "loss": 0.9023,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990128334561033e-05,
+      "loss": 1.3068,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989961754994896e-05,
+      "loss": 0.732,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998979378239453e-05,
+      "loss": 1.0073,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989624416783353e-05,
+      "loss": 0.9633,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989453658184987e-05,
+      "loss": 1.0563,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989281506623243e-05,
+      "loss": 0.7741,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989107962122123e-05,
+      "loss": 0.7091,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998893302470583e-05,
+      "loss": 0.8473,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998875669439876e-05,
+      "loss": 0.7885,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988578971225494e-05,
+      "loss": 1.227,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988399855210817e-05,
+      "loss": 0.8137,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988219346379712e-05,
+      "loss": 0.7638,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998803744475734e-05,
+      "loss": 0.7694,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998785415036907e-05,
+      "loss": 0.6659,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987669463240466e-05,
+      "loss": 1.2901,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987483383397275e-05,
+      "loss": 1.0722,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987295910865446e-05,
+      "loss": 0.8194,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998710704567112e-05,
+      "loss": 0.8583,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998691678784064e-05,
+      "loss": 0.7834,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986725137400526e-05,
+      "loss": 0.7683,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998653209437751e-05,
+      "loss": 0.9523,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986337658798503e-05,
+      "loss": 0.7375,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986141830690626e-05,
+      "loss": 0.8516,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985944610081184e-05,
+      "loss": 0.8718,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998574599699767e-05,
+      "loss": 1.0272,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998554599146779e-05,
+      "loss": 0.716,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985344593519432e-05,
+      "loss": 0.7085,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998514180318067e-05,
+      "loss": 0.7241,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984937620479793e-05,
+      "loss": 0.8099,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998473204544527e-05,
+      "loss": 0.7892,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984525078105762e-05,
+      "loss": 0.7748,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9984316718490136e-05,
+      "loss": 0.774,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998410696662744e-05,
+      "loss": 0.8209,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998389582254693e-05,
+      "loss": 0.8635,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998368328627804e-05,
+      "loss": 0.7746,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983469357850415e-05,
+      "loss": 0.7508,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983254037293878e-05,
+      "loss": 0.9812,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998303732463846e-05,
+      "loss": 0.7849,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998281921991438e-05,
+      "loss": 1.0265,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998259972315205e-05,
+      "loss": 0.6536,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982378834382073e-05,
+      "loss": 0.9035,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982156553635255e-05,
+      "loss": 0.805,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981932880942593e-05,
+      "loss": 0.6578,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981707816335275e-05,
+      "loss": 0.9811,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998148135984468e-05,
+      "loss": 0.913,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981253511502393e-05,
+      "loss": 0.8004,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981024271340184e-05,
+      "loss": 0.7934,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980793639390018e-05,
+      "loss": 1.2212,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980561615684054e-05,
+      "loss": 1.1539,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998032820025465e-05,
+      "loss": 0.7842,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998009339313435e-05,
+      "loss": 0.9906,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979857194355898e-05,
+      "loss": 1.1832,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997961960395223e-05,
+      "loss": 0.7656,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997938062195648e-05,
+      "loss": 1.2127,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979140248401966e-05,
+      "loss": 1.2169,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978898483322215e-05,
+      "loss": 0.8038,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978655326750928e-05,
+      "loss": 0.7835,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997841077872202e-05,
+      "loss": 1.0272,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978164839269588e-05,
+      "loss": 1.2233,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977917508427932e-05,
+      "loss": 0.9269,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977668786231536e-05,
+      "loss": 0.8743,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997741867271508e-05,
+      "loss": 0.8778,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997716716791345e-05,
+      "loss": 0.7818,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976914271861703e-05,
+      "loss": 0.8201,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976659984595115e-05,
+      "loss": 0.8189,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997640430614914e-05,
+      "loss": 0.7765,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976147236559435e-05,
+      "loss": 1.0314,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997588877586184e-05,
+      "loss": 0.7159,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975628924092405e-05,
+      "loss": 0.7005,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975367681287358e-05,
+      "loss": 1.0514,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975105047483124e-05,
+      "loss": 1.0068,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974841022716332e-05,
+      "loss": 0.7939,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974575607023795e-05,
+      "loss": 1.2433,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974308800442525e-05,
+      "loss": 0.8624,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974040603009726e-05,
+      "loss": 0.7272,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99737710147628e-05,
+      "loss": 0.8301,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973500035739333e-05,
+      "loss": 1.1319,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973227665977118e-05,
+      "loss": 0.9663,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972953905514126e-05,
+      "loss": 0.7773,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972678754388538e-05,
+      "loss": 1.0173,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972402212638725e-05,
+      "loss": 0.7291,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997212428030324e-05,
+      "loss": 0.7631,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971844957420844e-05,
+      "loss": 0.7232,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971564244030485e-05,
+      "loss": 0.9424,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997128214017131e-05,
+      "loss": 0.8339,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997099864588265e-05,
+      "loss": 1.0644,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970713761204045e-05,
+      "loss": 0.8283,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970427486175216e-05,
+      "loss": 0.748,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997013982083608e-05,
+      "loss": 0.877,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996985076522675e-05,
+      "loss": 1.1463,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996956031938754e-05,
+      "loss": 0.8442,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969268483358943e-05,
+      "loss": 1.0053,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968975257181656e-05,
+      "loss": 1.151,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968680640896566e-05,
+      "loss": 1.1698,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996838463454476e-05,
+      "loss": 0.7182,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968087238167512e-05,
+      "loss": 0.7547,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967788451806286e-05,
+      "loss": 0.874,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967488275502757e-05,
+      "loss": 0.8475,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967186709298772e-05,
+      "loss": 0.8256,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966883753236388e-05,
+      "loss": 0.8456,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966579407357852e-05,
+      "loss": 0.8462,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966273671705596e-05,
+      "loss": 1.0134,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996596654632226e-05,
+      "loss": 0.6877,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965658031250667e-05,
+      "loss": 0.7156,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965348126533835e-05,
+      "loss": 0.7482,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965036832214985e-05,
+      "loss": 0.9108,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996472414833752e-05,
+      "loss": 0.7143,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964410074945044e-05,
+      "loss": 1.2175,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964094612081347e-05,
+      "loss": 0.9799,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963777759790428e-05,
+      "loss": 0.7859,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963459518116458e-05,
+      "loss": 0.8351,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963139887103825e-05,
+      "loss": 1.0024,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962818866797096e-05,
+      "loss": 0.8112,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996249645724103e-05,
+      "loss": 0.7629,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996217265848059e-05,
+      "loss": 0.8839,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961847470560925e-05,
+      "loss": 0.9194,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961520893527385e-05,
+      "loss": 0.8225,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99611929274255e-05,
+      "loss": 1.1226,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9960863572301012e-05,
+      "loss": 0.688,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960532828199845e-05,
+      "loss": 1.1138,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9960200695168118e-05,
+      "loss": 0.8861,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959867173252137e-05,
+      "loss": 0.7446,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959532262498426e-05,
+      "loss": 0.9502,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995919596295367e-05,
+      "loss": 0.8295,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958858274664777e-05,
+      "loss": 1.1564,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958519197678825e-05,
+      "loss": 1.0209,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99581787320431e-05,
+      "loss": 0.851,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995783687780508e-05,
+      "loss": 0.6603,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957493635012433e-05,
+      "loss": 1.1516,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995714900371302e-05,
+      "loss": 1.0493,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99568029839549e-05,
+      "loss": 0.7652,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995645557578632e-05,
+      "loss": 0.8865,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9956106779255726e-05,
+      "loss": 1.1626,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955756594411753e-05,
+      "loss": 0.7883,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955405021303237e-05,
+      "loss": 0.7618,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99550520599792e-05,
+      "loss": 0.8345,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954697710488857e-05,
+      "loss": 0.7762,
+      "step": 516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954341972881625e-05,
+      "loss": 0.9631,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953984847207103e-05,
+      "loss": 0.8521,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953626333515096e-05,
+      "loss": 0.8367,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953266431855592e-05,
+      "loss": 0.8086,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995290514227878e-05,
+      "loss": 0.7797,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995254246483504e-05,
+      "loss": 0.8707,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952178399574943e-05,
+      "loss": 0.7207,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951812946549253e-05,
+      "loss": 0.888,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951446105808933e-05,
+      "loss": 1.0775,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995107787740514e-05,
+      "loss": 1.284,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950708261389215e-05,
+      "loss": 0.8827,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950337257812698e-05,
+      "loss": 0.6855,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994996486672733e-05,
+      "loss": 0.9378,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994959108818503e-05,
+      "loss": 0.8308,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949215922237926e-05,
+      "loss": 0.8645,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948839368938327e-05,
+      "loss": 0.749,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948461428338745e-05,
+      "loss": 1.2939,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948082100491884e-05,
+      "loss": 0.8174,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947701385450628e-05,
+      "loss": 1.0157,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9947319283268075e-05,
+      "loss": 1.1412,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946935793997503e-05,
+      "loss": 0.7431,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946550917692388e-05,
+      "loss": 0.7212,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9946164654406397e-05,
+      "loss": 0.7034,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994577700419339e-05,
+      "loss": 1.0609,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9945387967107428e-05,
+      "loss": 0.8373,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994499754320276e-05,
+      "loss": 0.7862,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994460573253382e-05,
+      "loss": 0.8544,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944212535155247e-05,
+      "loss": 0.7773,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943817951121875e-05,
+      "loss": 0.7717,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943421980488718e-05,
+      "loss": 0.8304,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943024623310998e-05,
+      "loss": 0.8233,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9942625879644122e-05,
+      "loss": 0.8333,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994222574954369e-05,
+      "loss": 0.8009,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941824233065498e-05,
+      "loss": 0.8673,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941421330265542e-05,
+      "loss": 0.7754,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941017041199992e-05,
+      "loss": 1.2773,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994061136592523e-05,
+      "loss": 1.0864,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940204304497825e-05,
+      "loss": 1.279,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939795856974537e-05,
+      "loss": 0.7549,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939386023412322e-05,
+      "loss": 0.7651,
+      "step": 556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993897480386833e-05,
+      "loss": 0.8599,
+      "step": 557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9938562198399906e-05,
+      "loss": 0.6591,
+      "step": 558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9938148207064576e-05,
+      "loss": 0.7381,
+      "step": 559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937732829920074e-05,
+      "loss": 1.0122,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937316067024325e-05,
+      "loss": 0.8058,
+      "step": 561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936897918435437e-05,
+      "loss": 0.7047,
+      "step": 562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993647838421172e-05,
+      "loss": 0.7016,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9936057464411674e-05,
+      "loss": 0.7577,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935635159094e-05,
+      "loss": 0.725,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993521146831758e-05,
+      "loss": 0.7885,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934786392141495e-05,
+      "loss": 0.8563,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993435993062502e-05,
+      "loss": 0.658,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933932083827625e-05,
+      "loss": 0.7779,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933502851808967e-05,
+      "loss": 1.1495,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9933072234628903e-05,
+      "loss": 1.2419,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932640232347476e-05,
+      "loss": 0.8028,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932206845024925e-05,
+      "loss": 0.8518,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931772072721683e-05,
+      "loss": 0.8703,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931335915498386e-05,
+      "loss": 1.1099,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930898373415837e-05,
+      "loss": 0.6731,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993045944653506e-05,
+      "loss": 1.1463,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993001913491726e-05,
+      "loss": 1.1379,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992957743862383e-05,
+      "loss": 0.9351,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929134357716367e-05,
+      "loss": 0.8107,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992868989225665e-05,
+      "loss": 1.126,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992824404230666e-05,
+      "loss": 0.7146,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992779680792857e-05,
+      "loss": 0.8364,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992734818918474e-05,
+      "loss": 0.9018,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992689818613773e-05,
+      "loss": 0.7452,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9926446798850293e-05,
+      "loss": 0.8878,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992599402738536e-05,
+      "loss": 0.892,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925539871806077e-05,
+      "loss": 0.7839,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992508433217577e-05,
+      "loss": 0.8224,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924627408557963e-05,
+      "loss": 0.8413,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992416910101637e-05,
+      "loss": 0.9486,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923709409614896e-05,
+      "loss": 0.9337,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9923248334417648e-05,
+      "loss": 0.7352,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992278587548891e-05,
+      "loss": 0.7325,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922322032893184e-05,
+      "loss": 1.0095,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9921856806695134e-05,
+      "loss": 0.7026,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992139019695964e-05,
+      "loss": 0.7452,
+      "step": 597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992092220375177e-05,
+      "loss": 1.1797,
+      "step": 598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992045282713678e-05,
+      "loss": 0.6866,
+      "step": 599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991998206718012e-05,
+      "loss": 0.8942,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919509923947434e-05,
+      "loss": 0.6842,
+      "step": 601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9919036397504565e-05,
+      "loss": 0.8158,
+      "step": 602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918561487917536e-05,
+      "loss": 0.8851,
+      "step": 603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9918085195252574e-05,
+      "loss": 0.9309,
+      "step": 604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917607519576095e-05,
+      "loss": 0.7687,
+      "step": 605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917128460954706e-05,
+      "loss": 0.5921,
+      "step": 606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916648019455208e-05,
+      "loss": 0.7409,
+      "step": 607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9916166195144598e-05,
+      "loss": 0.7917,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915682988090063e-05,
+      "loss": 0.9037,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9915198398358982e-05,
+      "loss": 0.7327,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914712426018924e-05,
+      "loss": 0.7881,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914225071137664e-05,
+      "loss": 0.876,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9913736333783152e-05,
+      "loss": 0.882,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991324621402354e-05,
+      "loss": 1.1941,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912754711927175e-05,
+      "loss": 0.8196,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912261827562593e-05,
+      "loss": 0.9535,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911767560998525e-05,
+      "loss": 0.8262,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9911271912303887e-05,
+      "loss": 0.7425,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910774881547803e-05,
+      "loss": 0.7432,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910276468799576e-05,
+      "loss": 0.9624,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909776674128703e-05,
+      "loss": 0.6834,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990927549760488e-05,
+      "loss": 0.8953,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908772939297995e-05,
+      "loss": 0.8398,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908268999278127e-05,
+      "loss": 0.8154,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990776367761554e-05,
+      "loss": 0.779,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907256974380703e-05,
+      "loss": 0.8374,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906748889644276e-05,
+      "loss": 0.8506,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906239423477103e-05,
+      "loss": 0.8656,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9905728575950225e-05,
+      "loss": 1.0405,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990521634713488e-05,
+      "loss": 0.6975,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990470273710249e-05,
+      "loss": 1.1748,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904187745924677e-05,
+      "loss": 1.0968,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903671373673258e-05,
+      "loss": 0.8779,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990315362042023e-05,
+      "loss": 0.8196,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902634486237794e-05,
+      "loss": 0.9387,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902113971198337e-05,
+      "loss": 0.7723,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901592075374447e-05,
+      "loss": 0.7252,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901068798838898e-05,
+      "loss": 1.211,
+      "step": 638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900544141664652e-05,
+      "loss": 1.0973,
+      "step": 639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990001810392487e-05,
+      "loss": 0.832,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9899490685692907e-05,
+      "loss": 0.7646,
+      "step": 641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989896188704231e-05,
+      "loss": 1.0668,
+      "step": 642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898431708046815e-05,
+      "loss": 0.8622,
+      "step": 643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989790014878035e-05,
+      "loss": 0.6081,
+      "step": 644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897367209317037e-05,
+      "loss": 0.742,
+      "step": 645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896832889731194e-05,
+      "loss": 0.6748,
+      "step": 646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896297190097327e-05,
+      "loss": 0.8677,
+      "step": 647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895760110490133e-05,
+      "loss": 0.8478,
+      "step": 648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9895221650984508e-05,
+      "loss": 0.7008,
+      "step": 649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894681811655538e-05,
+      "loss": 0.7966,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894140592578496e-05,
+      "loss": 0.9418,
+      "step": 651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893597993828847e-05,
+      "loss": 0.8534,
+      "step": 652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893054015482263e-05,
+      "loss": 0.8378,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892508657614596e-05,
+      "loss": 0.7801,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989196192030189e-05,
+      "loss": 1.1214,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989141380362038e-05,
+      "loss": 0.7083,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890864307646508e-05,
+      "loss": 0.9848,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890313432456887e-05,
+      "loss": 0.6918,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889761178128336e-05,
+      "loss": 0.8445,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889207544737864e-05,
+      "loss": 1.1194,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888652532362674e-05,
+      "loss": 0.7291,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888096141080157e-05,
+      "loss": 1.0784,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9887538370967898e-05,
+      "loss": 0.7184,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988697922210367e-05,
+      "loss": 1.1298,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988641869456545e-05,
+      "loss": 0.7834,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885856788431395e-05,
+      "loss": 0.9207,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988529350377986e-05,
+      "loss": 0.8188,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884728840689395e-05,
+      "loss": 1.0735,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988416279923873e-05,
+      "loss": 0.8074,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883595379506807e-05,
+      "loss": 0.8489,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988302658157274e-05,
+      "loss": 0.9003,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988245640551585e-05,
+      "loss": 0.7091,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988188485141564e-05,
+      "loss": 0.6862,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988131191935181e-05,
+      "loss": 0.8065,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880737609404256e-05,
+      "loss": 0.5603,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880161921653053e-05,
+      "loss": 0.9097,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879584856178485e-05,
+      "loss": 0.7284,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9879006413061018e-05,
+      "loss": 0.7809,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987842659238131e-05,
+      "loss": 0.7262,
+      "step": 679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877845394220214e-05,
+      "loss": 0.8826,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877262818658776e-05,
+      "loss": 0.7049,
+      "step": 681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987667886577823e-05,
+      "loss": 0.8365,
+      "step": 682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876093535660006e-05,
+      "loss": 1.009,
+      "step": 683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9875506828385723e-05,
+      "loss": 1.2526,
+      "step": 684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874918744037194e-05,
+      "loss": 1.1376,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874329282696423e-05,
+      "loss": 0.8525,
+      "step": 686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987373844444561e-05,
+      "loss": 1.155,
+      "step": 687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987314622936714e-05,
+      "loss": 0.8796,
+      "step": 688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872552637543592e-05,
+      "loss": 0.8792,
+      "step": 689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871957669057743e-05,
+      "loss": 0.8353,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871361323992553e-05,
+      "loss": 0.9605,
+      "step": 691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870763602431182e-05,
+      "loss": 0.7204,
+      "step": 692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870164504456973e-05,
+      "loss": 0.7573,
+      "step": 693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869564030153477e-05,
+      "loss": 0.7697,
+      "step": 694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986896217960441e-05,
+      "loss": 1.1686,
+      "step": 695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9868358952893715e-05,
+      "loss": 0.9463,
+      "step": 696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867754350105494e-05,
+      "loss": 0.774,
+      "step": 697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867148371324058e-05,
+      "loss": 0.8582,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986654101663391e-05,
+      "loss": 0.873,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865932286119737e-05,
+      "loss": 0.7345,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865322179866428e-05,
+      "loss": 0.7329,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9864710697959052e-05,
+      "loss": 1.1588,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9864097840482882e-05,
+      "loss": 0.7877,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986348360752337e-05,
+      "loss": 0.689,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862867999166176e-05,
+      "loss": 0.7853,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862251015497136e-05,
+      "loss": 1.077,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9861632656602286e-05,
+      "loss": 1.0209,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986101292256785e-05,
+      "loss": 0.6847,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986039181348025e-05,
+      "loss": 1.1262,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985976932942609e-05,
+      "loss": 0.6046,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985914547049218e-05,
+      "loss": 1.0308,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9858520236765503e-05,
+      "loss": 0.8251,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857893628333248e-05,
+      "loss": 1.2014,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9857265645282796e-05,
+      "loss": 0.7195,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9856636287701707e-05,
+      "loss": 0.8054,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9856005555677745e-05,
+      "loss": 1.1026,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9855373449298865e-05,
+      "loss": 1.1672,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854739968653202e-05,
+      "loss": 0.9792,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854105113829096e-05,
+      "loss": 0.8201,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985346888491507e-05,
+      "loss": 0.762,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9852831281999845e-05,
+      "loss": 0.8068,
+      "step": 721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985219230517233e-05,
+      "loss": 0.8222,
+      "step": 722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851551954521626e-05,
+      "loss": 1.0474,
+      "step": 723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9850910230137026e-05,
+      "loss": 1.1025,
+      "step": 724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985026713210801e-05,
+      "loss": 0.7183,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9849622660524262e-05,
+      "loss": 0.8814,
+      "step": 726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848976815475644e-05,
+      "loss": 0.7074,
+      "step": 727
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984832959705221e-05,
+      "loss": 0.9395,
+      "step": 728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984768100534422e-05,
+      "loss": 0.7744,
+      "step": 729
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847031040442113e-05,
+      "loss": 0.6945,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9846379702436518e-05,
+      "loss": 0.8075,
+      "step": 731
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845726991418265e-05,
+      "loss": 0.8841,
+      "step": 732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845072907478364e-05,
+      "loss": 0.7531,
+      "step": 733
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984441745070803e-05,
+      "loss": 0.8334,
+      "step": 734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843760621198658e-05,
+      "loss": 0.9015,
+      "step": 735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9843102419041838e-05,
+      "loss": 0.8178,
+      "step": 736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842442844329353e-05,
+      "loss": 0.7663,
+      "step": 737
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984178189715318e-05,
+      "loss": 0.742,
+      "step": 738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984111957760548e-05,
+      "loss": 0.9565,
+      "step": 739
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9840455885778604e-05,
+      "loss": 1.0998,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983979082176511e-05,
+      "loss": 0.8779,
+      "step": 741
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983912438565773e-05,
+      "loss": 1.0815,
+      "step": 742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9838456577549398e-05,
+      "loss": 0.6527,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983778739753323e-05,
+      "loss": 0.674,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983711684570254e-05,
+      "loss": 0.7568,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9836444922150837e-05,
+      "loss": 0.7826,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983577162697181e-05,
+      "loss": 0.8509,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9835096960259353e-05,
+      "loss": 0.7567,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983442092210754e-05,
+      "loss": 0.7334,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983374351261064e-05,
+      "loss": 0.674,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9833064731863107e-05,
+      "loss": 0.9169,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9832384579959603e-05,
+      "loss": 1.1012,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9831703056994963e-05,
+      "loss": 0.8753,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983102016306423e-05,
+      "loss": 0.7163,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983033589826262e-05,
+      "loss": 0.9091,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9829650262685548e-05,
+      "loss": 1.0776,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982896325642863e-05,
+      "loss": 0.8345,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982827487958766e-05,
+      "loss": 0.8339,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9827585132258627e-05,
+      "loss": 0.7296,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826894014537713e-05,
+      "loss": 0.706,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826201526521287e-05,
+      "loss": 0.8145,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982550766830592e-05,
+      "loss": 0.6876,
+      "step": 762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824812439988354e-05,
+      "loss": 0.9995,
+      "step": 763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982411584166554e-05,
+      "loss": 0.7722,
+      "step": 764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9823417873434616e-05,
+      "loss": 1.0748,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982271853539291e-05,
+      "loss": 1.081,
+      "step": 766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9822017827637933e-05,
+      "loss": 0.8053,
+      "step": 767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.98213157502674e-05,
+      "loss": 0.7625,
+      "step": 768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982061230337921e-05,
+      "loss": 0.787,
+      "step": 769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9819907487071452e-05,
+      "loss": 1.0447,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981920130144241e-05,
+      "loss": 0.9096,
+      "step": 771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818493746590558e-05,
+      "loss": 0.7619,
+      "step": 772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9817784822614558e-05,
+      "loss": 0.7339,
+      "step": 773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9817074529613262e-05,
+      "loss": 1.0592,
+      "step": 774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981636286768572e-05,
+      "loss": 0.8188,
+      "step": 775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9815649836931166e-05,
+      "loss": 0.9294,
+      "step": 776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981493543744903e-05,
+      "loss": 0.668,
+      "step": 777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981421966933893e-05,
+      "loss": 0.6867,
+      "step": 778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9813502532700672e-05,
+      "loss": 0.6301,
+      "step": 779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9812784027634257e-05,
+      "loss": 0.8258,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981206415423988e-05,
+      "loss": 0.6804,
+      "step": 781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981134291261792e-05,
+      "loss": 0.7917,
+      "step": 782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9810620302868944e-05,
+      "loss": 1.1271,
+      "step": 783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9809896325093722e-05,
+      "loss": 1.1686,
+      "step": 784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9809170979393207e-05,
+      "loss": 0.7159,
+      "step": 785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808444265868543e-05,
+      "loss": 0.8458,
+      "step": 786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9807716184621064e-05,
+      "loss": 0.7695,
+      "step": 787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806986735752293e-05,
+      "loss": 0.8426,
+      "step": 788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806255919363958e-05,
+      "loss": 0.7638,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9805523735557958e-05,
+      "loss": 0.7291,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9804790184436386e-05,
+      "loss": 0.7271,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9804055266101546e-05,
+      "loss": 0.9439,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9803318980655905e-05,
+      "loss": 1.0815,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9802581328202137e-05,
+      "loss": 0.8137,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.98018423088431e-05,
+      "loss": 0.8158,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9801101922681855e-05,
+      "loss": 0.9641,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980036016982163e-05,
+      "loss": 1.0434,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979961705036587e-05,
+      "loss": 0.6876,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9798872564418193e-05,
+      "loss": 0.727,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979812671208241e-05,
+      "loss": 0.816,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9797379493462527e-05,
+      "loss": 0.8263,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.979663090866274e-05,
+      "loss": 0.6945,
+      "step": 802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9795880957787435e-05,
+      "loss": 0.8167,
+      "step": 803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9795129640941187e-05,
+      "loss": 0.5619,
+      "step": 804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979437695822876e-05,
+      "loss": 0.7164,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979362290975511e-05,
+      "loss": 0.7019,
+      "step": 806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979286749562539e-05,
+      "loss": 0.7448,
+      "step": 807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9792110715944933e-05,
+      "loss": 0.6893,
+      "step": 808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9791352570819268e-05,
+      "loss": 0.898,
+      "step": 809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979059306035411e-05,
+      "loss": 0.7111,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789832184655374e-05,
+      "loss": 1.0047,
+      "step": 811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9789069943829154e-05,
+      "loss": 0.9598,
+      "step": 812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978830633798174e-05,
+      "loss": 1.1039,
+      "step": 813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9787541367219617e-05,
+      "loss": 0.9802,
+      "step": 814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9786775031649447e-05,
+      "loss": 0.7015,
+      "step": 815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97860073313781e-05,
+      "loss": 0.746,
+      "step": 816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9785238266512614e-05,
+      "loss": 0.9049,
+      "step": 817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9784467837160243e-05,
+      "loss": 0.5864,
+      "step": 818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978369604342841e-05,
+      "loss": 0.8277,
+      "step": 819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978292288542474e-05,
+      "loss": 0.8998,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978214836325704e-05,
+      "loss": 0.761,
+      "step": 821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9781372477033327e-05,
+      "loss": 0.6322,
+      "step": 822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978059522686177e-05,
+      "loss": 0.7348,
+      "step": 823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977981661285077e-05,
+      "loss": 0.7017,
+      "step": 824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9779036635108892e-05,
+      "loss": 0.8361,
+      "step": 825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97782552937449e-05,
+      "loss": 0.7715,
+      "step": 826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977747258886775e-05,
+      "loss": 0.756,
+      "step": 827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9776688520586575e-05,
+      "loss": 0.7684,
+      "step": 828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9775903089010718e-05,
+      "loss": 1.0814,
+      "step": 829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9775116294249702e-05,
+      "loss": 0.7762,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9774328136413234e-05,
+      "loss": 0.7235,
+      "step": 831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9773538615611224e-05,
+      "loss": 0.86,
+      "step": 832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977274773195376e-05,
+      "loss": 0.8258,
+      "step": 833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9771955485551127e-05,
+      "loss": 1.2438,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97711618765138e-05,
+      "loss": 0.9737,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9770366904952447e-05,
+      "loss": 0.7761,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9769570570977912e-05,
+      "loss": 1.1979,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9768772874701244e-05,
+      "loss": 0.8343,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9767973816233678e-05,
+      "loss": 0.8817,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976717339568663e-05,
+      "loss": 0.9023,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976637161317172e-05,
+      "loss": 0.7632,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976556846880075e-05,
+      "loss": 0.701,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764763962685714e-05,
+      "loss": 0.8531,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9763958094938794e-05,
+      "loss": 0.8154,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9763150865672362e-05,
+      "loss": 0.7182,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.976234227499898e-05,
+      "loss": 0.7689,
+      "step": 846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9761532323031405e-05,
+      "loss": 1.0308,
+      "step": 847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9760721009882577e-05,
+      "loss": 0.7403,
+      "step": 848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759908335665624e-05,
+      "loss": 0.7638,
+      "step": 849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759094300493878e-05,
+      "loss": 0.776,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975827890448084e-05,
+      "loss": 1.3343,
+      "step": 851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975746214774022e-05,
+      "loss": 1.0796,
+      "step": 852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9756644030385905e-05,
+      "loss": 0.8031,
+      "step": 853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9755824552531972e-05,
+      "loss": 0.7487,
+      "step": 854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97550037142927e-05,
+      "loss": 0.7523,
+      "step": 855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975418151578255e-05,
+      "loss": 0.7879,
+      "step": 856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975335795711616e-05,
+      "loss": 0.8495,
+      "step": 857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9752533038408385e-05,
+      "loss": 1.0691,
+      "step": 858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9751706759774248e-05,
+      "loss": 0.8001,
+      "step": 859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750879121328962e-05,
+      "loss": 0.8335,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750050123187944e-05,
+      "loss": 1.1259,
+      "step": 861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974921976546679e-05,
+      "loss": 1.0169,
+      "step": 862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9748388048281287e-05,
+      "loss": 0.7965,
+      "step": 863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9747554971747417e-05,
+      "loss": 0.8421,
+      "step": 864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9746720535981337e-05,
+      "loss": 0.7534,
+      "step": 865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745884741099415e-05,
+      "loss": 0.9279,
+      "step": 866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745047587218188e-05,
+      "loss": 0.8749,
+      "step": 867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9744209074454397e-05,
+      "loss": 1.1105,
+      "step": 868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743369202924967e-05,
+      "loss": 0.6997,
+      "step": 869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974252797274701e-05,
+      "loss": 1.075,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9741685384037827e-05,
+      "loss": 0.8674,
+      "step": 871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9740841436914917e-05,
+      "loss": 1.1663,
+      "step": 872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9739996131495963e-05,
+      "loss": 0.8745,
+      "step": 873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9739149467898835e-05,
+      "loss": 1.0082,
+      "step": 874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9738301446241595e-05,
+      "loss": 1.0991,
+      "step": 875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9737452066642496e-05,
+      "loss": 1.1648,
+      "step": 876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9736601329219975e-05,
+      "loss": 0.6774,
+      "step": 877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973574923409266e-05,
+      "loss": 0.7431,
+      "step": 878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973489578137938e-05,
+      "loss": 0.8513,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973404097119913e-05,
+      "loss": 0.8023,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9733184803671117e-05,
+      "loss": 0.8209,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9732327278914724e-05,
+      "loss": 0.6855,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973146839704953e-05,
+      "loss": 0.6443,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97306081581953e-05,
+      "loss": 0.6441,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9729746562471986e-05,
+      "loss": 0.9062,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728883609999735e-05,
+      "loss": 0.7793,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728019300898876e-05,
+      "loss": 0.7772,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9727153635289935e-05,
+      "loss": 0.8299,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9726286613293615e-05,
+      "loss": 0.8995,
+      "step": 889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972541823503083e-05,
+      "loss": 1.1036,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9724548500622664e-05,
+      "loss": 0.6856,
+      "step": 891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972367741019039e-05,
+      "loss": 0.7647,
+      "step": 892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9722804963855485e-05,
+      "loss": 0.648,
+      "step": 893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.97219311617396e-05,
+      "loss": 0.686,
+      "step": 894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972105600396458e-05,
+      "loss": 1.1757,
+      "step": 895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972017949065246e-05,
+      "loss": 0.9,
+      "step": 896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9719301621925477e-05,
+      "loss": 0.612,
+      "step": 897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9718422397906023e-05,
+      "loss": 0.943,
+      "step": 898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717541818716715e-05,
+      "loss": 1.177,
+      "step": 899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9716659884480337e-05,
+      "loss": 0.6633,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971577659531987e-05,
+      "loss": 0.7431,
+      "step": 901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9714891951358487e-05,
+      "loss": 0.6225,
+      "step": 902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971400595271954e-05,
+      "loss": 1.1671,
+      "step": 903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.971311859952658e-05,
+      "loss": 0.6629,
+      "step": 904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712229891903342e-05,
+      "loss": 0.7176,
+      "step": 905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9711339829973747e-05,
+      "loss": 0.7351,
+      "step": 906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710448413861912e-05,
+      "loss": 0.7139,
+      "step": 907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9709555643692136e-05,
+      "loss": 0.7842,
+      "step": 908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9708661519588915e-05,
+      "loss": 0.8622,
+      "step": 909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9707766041676923e-05,
+      "loss": 0.8125,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970686921008103e-05,
+      "loss": 1.1351,
+      "step": 911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9705971024926294e-05,
+      "loss": 1.0143,
+      "step": 912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9705071486337962e-05,
+      "loss": 0.6383,
+      "step": 913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970417059444147e-05,
+      "loss": 0.6299,
+      "step": 914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9703268349362433e-05,
+      "loss": 0.8529,
+      "step": 915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9702364751226675e-05,
+      "loss": 0.9073,
+      "step": 916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970145980016019e-05,
+      "loss": 1.0448,
+      "step": 917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9700553496289167e-05,
+      "loss": 1.1009,
+      "step": 918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9699645839739987e-05,
+      "loss": 0.9728,
+      "step": 919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9698736830639214e-05,
+      "loss": 1.0084,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9697826469113606e-05,
+      "loss": 1.1134,
+      "step": 921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9696914755290107e-05,
+      "loss": 0.673,
+      "step": 922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9696001689295845e-05,
+      "loss": 1.202,
+      "step": 923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9695087271258146e-05,
+      "loss": 0.7336,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9694171501304513e-05,
+      "loss": 0.8164,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9693254379562654e-05,
+      "loss": 0.9729,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9692335906160448e-05,
+      "loss": 0.9233,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9691416081225973e-05,
+      "loss": 0.8941,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9690494904887494e-05,
+      "loss": 0.7336,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9689572377273457e-05,
+      "loss": 0.9036,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9688648498512508e-05,
+      "loss": 0.7684,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9687723268733472e-05,
+      "loss": 0.9526,
+      "step": 932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9686796688065373e-05,
+      "loss": 0.8085,
+      "step": 933
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9685868756637408e-05,
+      "loss": 1.2049,
+      "step": 934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9684939474578976e-05,
+      "loss": 0.7128,
+      "step": 935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9684008842019656e-05,
+      "loss": 1.1888,
+      "step": 936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968307685908922e-05,
+      "loss": 0.7519,
+      "step": 937
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968214352591763e-05,
+      "loss": 0.7329,
+      "step": 938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968120884263503e-05,
+      "loss": 1.1304,
+      "step": 939
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9680272809371757e-05,
+      "loss": 0.8493,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967933542625833e-05,
+      "loss": 0.7607,
+      "step": 941
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9678396693425464e-05,
+      "loss": 0.7713,
+      "step": 942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9677456611004064e-05,
+      "loss": 0.9783,
+      "step": 943
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967651517912521e-05,
+      "loss": 0.8393,
+      "step": 944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9675572397920187e-05,
+      "loss": 0.7832,
+      "step": 945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967462826752045e-05,
+      "loss": 0.7769,
+      "step": 946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967368278805766e-05,
+      "loss": 1.2433,
+      "step": 947
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9672735959663655e-05,
+      "loss": 0.6652,
+      "step": 948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.967178778247046e-05,
+      "loss": 0.5992,
+      "step": 949
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9670838256610297e-05,
+      "loss": 0.7327,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966988738221557e-05,
+      "loss": 1.027,
+      "step": 951
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966893515941887e-05,
+      "loss": 1.1668,
+      "step": 952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966798158835298e-05,
+      "loss": 0.6923,
+      "step": 953
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9667026669150876e-05,
+      "loss": 0.7365,
+      "step": 954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96660704019457e-05,
+      "loss": 0.9249,
+      "step": 955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966511278687081e-05,
+      "loss": 1.0178,
+      "step": 956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9664153824059732e-05,
+      "loss": 0.8868,
+      "step": 957
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9663193513646193e-05,
+      "loss": 0.7974,
+      "step": 958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9662231855764096e-05,
+      "loss": 0.8912,
+      "step": 959
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966126885054754e-05,
+      "loss": 1.2299,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660304498130813e-05,
+      "loss": 0.7174,
+      "step": 961
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659338798648383e-05,
+      "loss": 0.6507,
+      "step": 962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658371752234912e-05,
+      "loss": 1.2387,
+      "step": 963
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965740335902525e-05,
+      "loss": 1.0388,
+      "step": 964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965643361915443e-05,
+      "loss": 0.8515,
+      "step": 965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9655462532757677e-05,
+      "loss": 0.826,
+      "step": 966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9654490099970402e-05,
+      "loss": 0.8709,
+      "step": 967
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9653516320928208e-05,
+      "loss": 0.7395,
+      "step": 968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9652541195766876e-05,
+      "loss": 0.8592,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9651564724622387e-05,
+      "loss": 0.7542,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96505869076309e-05,
+      "loss": 0.5577,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9649607744928765e-05,
+      "loss": 1.0594,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964862723665252e-05,
+      "loss": 0.6335,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964764538293889e-05,
+      "loss": 0.7541,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964666218392479e-05,
+      "loss": 0.6932,
+      "step": 975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964567763974732e-05,
+      "loss": 1.1022,
+      "step": 976
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964469175054377e-05,
+      "loss": 0.6778,
+      "step": 977
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9643704516451608e-05,
+      "loss": 0.7029,
+      "step": 978
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9642715937608506e-05,
+      "loss": 0.7116,
+      "step": 979
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9641726014152308e-05,
+      "loss": 0.8738,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964073474622106e-05,
+      "loss": 0.8313,
+      "step": 981
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963974213395298e-05,
+      "loss": 0.7772,
+      "step": 982
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9638748177486488e-05,
+      "loss": 0.7146,
+      "step": 983
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963775287696018e-05,
+      "loss": 1.0181,
+      "step": 984
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9636756232512842e-05,
+      "loss": 0.6607,
+      "step": 985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963575824428346e-05,
+      "loss": 0.6964,
+      "step": 986
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9634758912411184e-05,
+      "loss": 0.7119,
+      "step": 987
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963375823703537e-05,
+      "loss": 0.7706,
+      "step": 988
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963275621829556e-05,
+      "loss": 0.7149,
+      "step": 989
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963175285633147e-05,
+      "loss": 0.9588,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963074815128302e-05,
+      "loss": 0.6964,
+      "step": 991
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962974210329031e-05,
+      "loss": 0.8806,
+      "step": 992
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9628734712493616e-05,
+      "loss": 0.8373,
+      "step": 993
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9627725979033424e-05,
+      "loss": 0.8082,
+      "step": 994
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9626715903050388e-05,
+      "loss": 1.0558,
+      "step": 995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9625704484685356e-05,
+      "loss": 0.7832,
+      "step": 996
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962469172407937e-05,
+      "loss": 0.6466,
+      "step": 997
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623677621373643e-05,
+      "loss": 0.7385,
+      "step": 998
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9622662176709596e-05,
+      "loss": 0.9861,
+      "step": 999
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9621645390228823e-05,
+      "loss": 0.6045,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9620627262073106e-05,
+      "loss": 0.6933,
+      "step": 1001
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961960779238441e-05,
+      "loss": 1.0736,
+      "step": 1002
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961858698130491e-05,
+      "loss": 0.6485,
+      "step": 1003
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961756482897693e-05,
+      "loss": 0.7974,
+      "step": 1004
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961654133554302e-05,
+      "loss": 0.6709,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961551650114589e-05,
+      "loss": 0.7641,
+      "step": 1006
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9614490325928452e-05,
+      "loss": 0.7463,
+      "step": 1007
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9613462810033797e-05,
+      "loss": 0.8236,
+      "step": 1008
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9612433953605202e-05,
+      "loss": 0.7313,
+      "step": 1009
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611403756786142e-05,
+      "loss": 0.7628,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961037221972026e-05,
+      "loss": 0.7928,
+      "step": 1011
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960933934255141e-05,
+      "loss": 0.8165,
+      "step": 1012
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9608305125423608e-05,
+      "loss": 0.7392,
+      "step": 1013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9607269568481077e-05,
+      "loss": 0.7785,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9606232671868214e-05,
+      "loss": 0.7728,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9605194435729608e-05,
+      "loss": 1.0154,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9604154860210037e-05,
+      "loss": 0.8642,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9603113945454463e-05,
+      "loss": 0.6618,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960207169160803e-05,
+      "loss": 0.6527,
+      "step": 1019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601028098816077e-05,
+      "loss": 0.7404,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9599983167224127e-05,
+      "loss": 0.6351,
+      "step": 1021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9598936896977884e-05,
+      "loss": 0.8753,
+      "step": 1022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959788928822325e-05,
+      "loss": 1.1321,
+      "step": 1023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95968403411063e-05,
+      "loss": 0.7994,
+      "step": 1024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959579005577331e-05,
+      "loss": 0.8044,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9594738432370734e-05,
+      "loss": 0.9652,
+      "step": 1026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9593685471045204e-05,
+      "loss": 0.6823,
+      "step": 1027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9592631171943562e-05,
+      "loss": 1.0278,
+      "step": 1028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959157553521282e-05,
+      "loss": 0.8992,
+      "step": 1029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9590518561000174e-05,
+      "loss": 0.9532,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9589460249453012e-05,
+      "loss": 0.7634,
+      "step": 1031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9588400600718914e-05,
+      "loss": 0.6496,
+      "step": 1032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958733961494564e-05,
+      "loss": 0.8894,
+      "step": 1033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9586277292281136e-05,
+      "loss": 0.7174,
+      "step": 1034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9585213632873534e-05,
+      "loss": 0.683,
+      "step": 1035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9584148636871157e-05,
+      "loss": 0.8924,
+      "step": 1036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9583082304422513e-05,
+      "loss": 0.7144,
+      "step": 1037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958201463567629e-05,
+      "loss": 0.6901,
+      "step": 1038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958094563078137e-05,
+      "loss": 0.8683,
+      "step": 1039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957987528988682e-05,
+      "loss": 0.7149,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9578803613141893e-05,
+      "loss": 0.7249,
+      "step": 1041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9577730600696026e-05,
+      "loss": 0.6857,
+      "step": 1042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957665625269884e-05,
+      "loss": 0.7959,
+      "step": 1043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957558056930015e-05,
+      "loss": 0.8626,
+      "step": 1044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957450355064995e-05,
+      "loss": 0.8591,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9573425196898426e-05,
+      "loss": 1.0782,
+      "step": 1046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9572345508195945e-05,
+      "loss": 0.8067,
+      "step": 1047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9571264484693064e-05,
+      "loss": 0.8201,
+      "step": 1048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9570182126540524e-05,
+      "loss": 0.6716,
+      "step": 1049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9569098433889253e-05,
+      "loss": 0.7525,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956801340689036e-05,
+      "loss": 0.7501,
+      "step": 1051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566927045695152e-05,
+      "loss": 0.7805,
+      "step": 1052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9565839350455116e-05,
+      "loss": 0.7412,
+      "step": 1053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9564750321321914e-05,
+      "loss": 0.8721,
+      "step": 1054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9563659958447414e-05,
+      "loss": 0.7296,
+      "step": 1055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9562568261983656e-05,
+      "loss": 1.0345,
+      "step": 1056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9561475232082868e-05,
+      "loss": 1.0002,
+      "step": 1057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9560380868897468e-05,
+      "loss": 0.7437,
+      "step": 1058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955928517258006e-05,
+      "loss": 0.6671,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9558188143283425e-05,
+      "loss": 0.8299,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955708978116054e-05,
+      "loss": 0.8295,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955599008636457e-05,
+      "loss": 0.9054,
+      "step": 1062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955488905904885e-05,
+      "loss": 0.7087,
+      "step": 1063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955378669936692e-05,
+      "loss": 0.927,
+      "step": 1064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9552683007472488e-05,
+      "loss": 0.978,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9551577983519464e-05,
+      "loss": 0.6783,
+      "step": 1066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550471627661937e-05,
+      "loss": 0.6604,
+      "step": 1067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9549363940054175e-05,
+      "loss": 0.7615,
+      "step": 1068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9548254920850637e-05,
+      "loss": 0.696,
+      "step": 1069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9547144570205973e-05,
+      "loss": 0.6761,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546032888275015e-05,
+      "loss": 0.7978,
+      "step": 1071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9544919875212776e-05,
+      "loss": 0.8552,
+      "step": 1072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954380553117446e-05,
+      "loss": 0.9153,
+      "step": 1073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542689856315458e-05,
+      "loss": 0.6123,
+      "step": 1074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9541572850791334e-05,
+      "loss": 0.7704,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540454514757858e-05,
+      "loss": 0.9327,
+      "step": 1076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953933484837097e-05,
+      "loss": 0.7138,
+      "step": 1077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95382138517868e-05,
+      "loss": 0.8433,
+      "step": 1078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953709152516166e-05,
+      "loss": 0.7311,
+      "step": 1079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9535967868652058e-05,
+      "loss": 0.7298,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9534842882414672e-05,
+      "loss": 0.8781,
+      "step": 1081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533716566606386e-05,
+      "loss": 0.6775,
+      "step": 1082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532588921384246e-05,
+      "loss": 0.9174,
+      "step": 1083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531459946905503e-05,
+      "loss": 0.6462,
+      "step": 1084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530329643327576e-05,
+      "loss": 0.7862,
+      "step": 1085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529198010808087e-05,
+      "loss": 0.9597,
+      "step": 1086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528065049504828e-05,
+      "loss": 0.6245,
+      "step": 1087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9526930759575786e-05,
+      "loss": 0.6837,
+      "step": 1088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525795141179133e-05,
+      "loss": 0.8168,
+      "step": 1089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952465819447322e-05,
+      "loss": 1.1408,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952351991961659e-05,
+      "loss": 0.8653,
+      "step": 1091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952238031676796e-05,
+      "loss": 0.8342,
+      "step": 1092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952123938608625e-05,
+      "loss": 0.8546,
+      "step": 1093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520097127730547e-05,
+      "loss": 1.2142,
+      "step": 1094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518953541860138e-05,
+      "loss": 0.8406,
+      "step": 1095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9517808628634484e-05,
+      "loss": 0.7024,
+      "step": 1096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516662388213237e-05,
+      "loss": 0.7493,
+      "step": 1097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515514820756235e-05,
+      "loss": 0.7657,
+      "step": 1098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9514365926423494e-05,
+      "loss": 0.8451,
+      "step": 1099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513215705375224e-05,
+      "loss": 0.7506,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512064157771817e-05,
+      "loss": 0.7615,
+      "step": 1101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510911283773842e-05,
+      "loss": 0.798,
+      "step": 1102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509757083542063e-05,
+      "loss": 0.573,
+      "step": 1103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508601557237427e-05,
+      "loss": 0.6252,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507444705021063e-05,
+      "loss": 1.1726,
+      "step": 1105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950628652705429e-05,
+      "loss": 0.738,
+      "step": 1106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9505127023498603e-05,
+      "loss": 0.7374,
+      "step": 1107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950396619451569e-05,
+      "loss": 0.7163,
+      "step": 1108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9502804040267418e-05,
+      "loss": 0.7278,
+      "step": 1109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9501640560915845e-05,
+      "loss": 0.8363,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950047575662321e-05,
+      "loss": 0.8556,
+      "step": 1111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9499309627551938e-05,
+      "loss": 0.5844,
+      "step": 1112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9498142173864636e-05,
+      "loss": 1.232,
+      "step": 1113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94969733957241e-05,
+      "loss": 0.6896,
+      "step": 1114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9495803293293305e-05,
+      "loss": 1.0999,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9494631866735418e-05,
+      "loss": 0.7478,
+      "step": 1116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949345911621378e-05,
+      "loss": 0.7737,
+      "step": 1117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9492285041891934e-05,
+      "loss": 0.651,
+      "step": 1118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9491109643933588e-05,
+      "loss": 0.5803,
+      "step": 1119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9489932922502646e-05,
+      "loss": 0.6224,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9488754877763197e-05,
+      "loss": 0.7521,
+      "step": 1121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9487575509879507e-05,
+      "loss": 0.718,
+      "step": 1122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9486394819016032e-05,
+      "loss": 1.014,
+      "step": 1123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9485212805337416e-05,
+      "loss": 0.6434,
+      "step": 1124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9484029469008476e-05,
+      "loss": 0.777,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9482844810194227e-05,
+      "loss": 0.6791,
+      "step": 1126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948165882905986e-05,
+      "loss": 0.7515,
+      "step": 1127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9480471525770747e-05,
+      "loss": 1.0142,
+      "step": 1128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947928290049245e-05,
+      "loss": 0.6342,
+      "step": 1129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9478092953390728e-05,
+      "loss": 0.871,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94769016846315e-05,
+      "loss": 0.7397,
+      "step": 1131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947570909438088e-05,
+      "loss": 0.6062,
+      "step": 1132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947451518280517e-05,
+      "loss": 0.7332,
+      "step": 1133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9473319950070854e-05,
+      "loss": 0.8074,
+      "step": 1134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472123396344598e-05,
+      "loss": 0.6538,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947092552179325e-05,
+      "loss": 0.7086,
+      "step": 1136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946972632658385e-05,
+      "loss": 0.721,
+      "step": 1137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946852581088362e-05,
+      "loss": 1.1403,
+      "step": 1138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946732397485996e-05,
+      "loss": 0.9673,
+      "step": 1139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946612081868046e-05,
+      "loss": 1.0079,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9464916342512884e-05,
+      "loss": 0.7033,
+      "step": 1141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9463710546525202e-05,
+      "loss": 0.7062,
+      "step": 1142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9462503430885545e-05,
+      "loss": 0.7362,
+      "step": 1143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9461294995762246e-05,
+      "loss": 1.0993,
+      "step": 1144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94600852413238e-05,
+      "loss": 0.8061,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945887416773891e-05,
+      "loss": 0.7716,
+      "step": 1146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9457661775176447e-05,
+      "loss": 0.7835,
+      "step": 1147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9456448063805474e-05,
+      "loss": 0.78,
+      "step": 1148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9455233033795236e-05,
+      "loss": 0.7677,
+      "step": 1149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9454016685315158e-05,
+      "loss": 0.8956,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945279901853485e-05,
+      "loss": 0.7895,
+      "step": 1151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9451580033624112e-05,
+      "loss": 0.6776,
+      "step": 1152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9450359730752922e-05,
+      "loss": 0.7721,
+      "step": 1153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9449138110091444e-05,
+      "loss": 1.1091,
+      "step": 1154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944791517181002e-05,
+      "loss": 0.8203,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944669091607919e-05,
+      "loss": 1.1165,
+      "step": 1156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9445465343069658e-05,
+      "loss": 0.8252,
+      "step": 1157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9444238452952332e-05,
+      "loss": 0.8179,
+      "step": 1158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9443010245898283e-05,
+      "loss": 0.6973,
+      "step": 1159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441780722078785e-05,
+      "loss": 0.7445,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9440549881665284e-05,
+      "loss": 0.7933,
+      "step": 1161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9439317724829412e-05,
+      "loss": 0.7006,
+      "step": 1162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9438084251742985e-05,
+      "loss": 0.9844,
+      "step": 1163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9436849462578005e-05,
+      "loss": 0.7864,
+      "step": 1164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943561335750665e-05,
+      "loss": 0.7153,
+      "step": 1165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9434375936701295e-05,
+      "loss": 0.8363,
+      "step": 1166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9433137200334483e-05,
+      "loss": 0.637,
+      "step": 1167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431897148578952e-05,
+      "loss": 0.7988,
+      "step": 1168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9430655781607615e-05,
+      "loss": 0.9489,
+      "step": 1169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9429413099593578e-05,
+      "loss": 0.7663,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9428169102710117e-05,
+      "loss": 0.8057,
+      "step": 1171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9426923791130706e-05,
+      "loss": 0.8678,
+      "step": 1172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9425677165028987e-05,
+      "loss": 0.9257,
+      "step": 1173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9424429224578808e-05,
+      "loss": 0.8917,
+      "step": 1174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9423179969954175e-05,
+      "loss": 1.0689,
+      "step": 1175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942192940132929e-05,
+      "loss": 1.0974,
+      "step": 1176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9420677518878535e-05,
+      "loss": 0.6868,
+      "step": 1177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9419424322776485e-05,
+      "loss": 0.6342,
+      "step": 1178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9418169813197886e-05,
+      "loss": 0.6404,
+      "step": 1179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9416913990317663e-05,
+      "loss": 1.1231,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941565685431094e-05,
+      "loss": 0.922,
+      "step": 1181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9414398405353017e-05,
+      "loss": 0.783,
+      "step": 1182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9413138643619374e-05,
+      "loss": 0.7126,
+      "step": 1183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9411877569285677e-05,
+      "loss": 0.8338,
+      "step": 1184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941061518252777e-05,
+      "loss": 0.9221,
+      "step": 1185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9409351483521694e-05,
+      "loss": 0.8269,
+      "step": 1186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408086472443658e-05,
+      "loss": 0.6858,
+      "step": 1187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940682014947006e-05,
+      "loss": 0.6344,
+      "step": 1188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405552514777475e-05,
+      "loss": 0.7149,
+      "step": 1189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9404283568542673e-05,
+      "loss": 0.7257,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9403013310942603e-05,
+      "loss": 0.8733,
+      "step": 1191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9401741742154385e-05,
+      "loss": 0.7722,
+      "step": 1192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9400468862355337e-05,
+      "loss": 0.6492,
+      "step": 1193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9399194671722954e-05,
+      "loss": 1.1039,
+      "step": 1194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9397919170434913e-05,
+      "loss": 0.6129,
+      "step": 1195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9396642358669066e-05,
+      "loss": 0.7654,
+      "step": 1196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939536423660347e-05,
+      "loss": 0.7884,
+      "step": 1197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939408480441634e-05,
+      "loss": 0.7849,
+      "step": 1198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939280406228609e-05,
+      "loss": 1.201,
+      "step": 1199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9391522010391304e-05,
+      "loss": 0.6841,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9390238648910765e-05,
+      "loss": 0.6308,
+      "step": 1201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9388953978023423e-05,
+      "loss": 0.7661,
+      "step": 1202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9387667997908418e-05,
+      "loss": 0.8726,
+      "step": 1203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9386380708745078e-05,
+      "loss": 0.7679,
+      "step": 1204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9385092110712898e-05,
+      "loss": 0.8587,
+      "step": 1205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9383802203991562e-05,
+      "loss": 0.7563,
+      "step": 1206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9382510988760953e-05,
+      "loss": 0.6834,
+      "step": 1207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.938121846520111e-05,
+      "loss": 1.2051,
+      "step": 1208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937992463349227e-05,
+      "loss": 1.0493,
+      "step": 1209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378629493814852e-05,
+      "loss": 0.8939,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9377333046349455e-05,
+      "loss": 0.8454,
+      "step": 1211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9376035291276854e-05,
+      "loss": 0.7257,
+      "step": 1212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937473622877802e-05,
+      "loss": 1.1543,
+      "step": 1213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9373435859034092e-05,
+      "loss": 0.7934,
+      "step": 1214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9372134182226403e-05,
+      "loss": 0.6112,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937083119853646e-05,
+      "loss": 0.8813,
+      "step": 1216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936952690814596e-05,
+      "loss": 1.0412,
+      "step": 1217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368221311236774e-05,
+      "loss": 0.7147,
+      "step": 1218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9366914407990953e-05,
+      "loss": 0.6841,
+      "step": 1219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936560619859075e-05,
+      "loss": 0.7575,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9364296683218574e-05,
+      "loss": 0.8713,
+      "step": 1221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9362985862057035e-05,
+      "loss": 0.7271,
+      "step": 1222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9361673735288913e-05,
+      "loss": 0.8155,
+      "step": 1223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936036030309718e-05,
+      "loss": 0.987,
+      "step": 1224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9359045565664986e-05,
+      "loss": 0.8099,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935772952317566e-05,
+      "loss": 0.681,
+      "step": 1226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9356412175812713e-05,
+      "loss": 0.8012,
+      "step": 1227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9355093523759846e-05,
+      "loss": 0.7502,
+      "step": 1228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9353773567200936e-05,
+      "loss": 1.0775,
+      "step": 1229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9352452306320036e-05,
+      "loss": 0.754,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935112974130139e-05,
+      "loss": 0.8778,
+      "step": 1231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9349805872329426e-05,
+      "loss": 0.7051,
+      "step": 1232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934848069958874e-05,
+      "loss": 0.7258,
+      "step": 1233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.934715422326413e-05,
+      "loss": 0.7006,
+      "step": 1234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9345826443540554e-05,
+      "loss": 0.6989,
+      "step": 1235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9344497360603162e-05,
+      "loss": 0.6551,
+      "step": 1236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9343166974637294e-05,
+      "loss": 0.9603,
+      "step": 1237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934183528582846e-05,
+      "loss": 0.8142,
+      "step": 1238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9340502294362355e-05,
+      "loss": 0.7116,
+      "step": 1239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933916800042485e-05,
+      "loss": 0.6148,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9337832404202012e-05,
+      "loss": 0.9616,
+      "step": 1241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9336495505880077e-05,
+      "loss": 0.6926,
+      "step": 1242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9335157305645473e-05,
+      "loss": 1.0509,
+      "step": 1243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933381780368479e-05,
+      "loss": 0.7905,
+      "step": 1244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9332477000184825e-05,
+      "loss": 1.1419,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331134895332536e-05,
+      "loss": 0.7167,
+      "step": 1246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932979148931508e-05,
+      "loss": 0.7572,
+      "step": 1247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932844678231977e-05,
+      "loss": 0.6905,
+      "step": 1248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9327100774534137e-05,
+      "loss": 0.7348,
+      "step": 1249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9325753466145862e-05,
+      "loss": 0.7372,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9324404857342814e-05,
+      "loss": 0.8477,
+      "step": 1251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9323054948313055e-05,
+      "loss": 0.6819,
+      "step": 1252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9321703739244816e-05,
+      "loss": 1.1826,
+      "step": 1253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9320351230326522e-05,
+      "loss": 0.6223,
+      "step": 1254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9318997421746762e-05,
+      "loss": 0.6515,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931764231369432e-05,
+      "loss": 0.7266,
+      "step": 1256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9316285906358158e-05,
+      "loss": 0.7492,
+      "step": 1257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931492819992742e-05,
+      "loss": 1.0459,
+      "step": 1258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931356919459142e-05,
+      "loss": 0.7688,
+      "step": 1259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931220889053967e-05,
+      "loss": 0.8169,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9310847287961854e-05,
+      "loss": 0.7392,
+      "step": 1261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9309484387047838e-05,
+      "loss": 0.7608,
+      "step": 1262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930812018798767e-05,
+      "loss": 0.6993,
+      "step": 1263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9306754690971582e-05,
+      "loss": 0.6591,
+      "step": 1264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9305387896189973e-05,
+      "loss": 0.7809,
+      "step": 1265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9304019803833446e-05,
+      "loss": 1.0446,
+      "step": 1266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9302650414092765e-05,
+      "loss": 0.7653,
+      "step": 1267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9301279727158883e-05,
+      "loss": 0.7341,
+      "step": 1268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299907743222938e-05,
+      "loss": 0.7415,
+      "step": 1269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9298534462476236e-05,
+      "loss": 0.7526,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9297159885110277e-05,
+      "loss": 1.1884,
+      "step": 1271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9295784011316735e-05,
+      "loss": 0.6923,
+      "step": 1272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929440684128747e-05,
+      "loss": 1.2045,
+      "step": 1273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.929302837521452e-05,
+      "loss": 0.7845,
+      "step": 1274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9291648613290097e-05,
+      "loss": 0.8136,
+      "step": 1275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9290267555706604e-05,
+      "loss": 0.8308,
+      "step": 1276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9288885202656617e-05,
+      "loss": 0.6628,
+      "step": 1277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9287501554332897e-05,
+      "loss": 0.6767,
+      "step": 1278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9286116610928387e-05,
+      "loss": 0.7052,
+      "step": 1279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9284730372636208e-05,
+      "loss": 0.648,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.928334283964966e-05,
+      "loss": 0.6717,
+      "step": 1281
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9281954012162224e-05,
+      "loss": 0.6319,
+      "step": 1282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9280563890367568e-05,
+      "loss": 0.7122,
+      "step": 1283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9279172474459532e-05,
+      "loss": 0.8196,
+      "step": 1284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927777976463214e-05,
+      "loss": 0.6899,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927638576107959e-05,
+      "loss": 0.8528,
+      "step": 1286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927499046399628e-05,
+      "loss": 0.7252,
+      "step": 1287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9273593873576763e-05,
+      "loss": 1.0048,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9272195990015793e-05,
+      "loss": 1.1579,
+      "step": 1289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927079681350829e-05,
+      "loss": 0.7128,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926939634424936e-05,
+      "loss": 0.6945,
+      "step": 1291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926799458243429e-05,
+      "loss": 0.8672,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926659152825855e-05,
+      "loss": 1.2099,
+      "step": 1293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265187181917782e-05,
+      "loss": 0.6506,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9263781543607817e-05,
+      "loss": 0.8321,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9262374613524655e-05,
+      "loss": 0.665,
+      "step": 1296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926096639186449e-05,
+      "loss": 0.8857,
+      "step": 1297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925955687882369e-05,
+      "loss": 0.8651,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9258146074598795e-05,
+      "loss": 1.0312,
+      "step": 1299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925673397938654e-05,
+      "loss": 0.7241,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9255320593383825e-05,
+      "loss": 0.6632,
+      "step": 1301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9253905916787743e-05,
+      "loss": 0.6559,
+      "step": 1302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.925248994979556e-05,
+      "loss": 0.7698,
+      "step": 1303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9251072692604725e-05,
+      "loss": 0.711,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924965414541286e-05,
+      "loss": 0.7841,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9248234308417774e-05,
+      "loss": 0.6637,
+      "step": 1306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9246813181817457e-05,
+      "loss": 0.734,
+      "step": 1307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9245390765810073e-05,
+      "loss": 1.0163,
+      "step": 1308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924396706059397e-05,
+      "loss": 1.0759,
+      "step": 1309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9242542066367673e-05,
+      "loss": 0.7477,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924111578332989e-05,
+      "loss": 0.7666,
+      "step": 1311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9239688211679502e-05,
+      "loss": 0.7851,
+      "step": 1312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923825935161558e-05,
+      "loss": 0.7376,
+      "step": 1313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923682920333737e-05,
+      "loss": 0.897,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9235397767044286e-05,
+      "loss": 0.8603,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9233965042935944e-05,
+      "loss": 0.8183,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9232531031212125e-05,
+      "loss": 0.7134,
+      "step": 1317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9231095732072788e-05,
+      "loss": 0.8016,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.922965914571808e-05,
+      "loss": 0.6811,
+      "step": 1319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9228221272348327e-05,
+      "loss": 0.773,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9226782112164023e-05,
+      "loss": 0.7196,
+      "step": 1321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9225341665365853e-05,
+      "loss": 0.7346,
+      "step": 1322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9223899932154678e-05,
+      "loss": 0.8091,
+      "step": 1323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9222456912731537e-05,
+      "loss": 0.7938,
+      "step": 1324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9221012607297657e-05,
+      "loss": 0.7257,
+      "step": 1325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921956701605442e-05,
+      "loss": 0.7213,
+      "step": 1326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9218120139203424e-05,
+      "loss": 0.8214,
+      "step": 1327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921667197694641e-05,
+      "loss": 0.7951,
+      "step": 1328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921522252948533e-05,
+      "loss": 0.8471,
+      "step": 1329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9213771797022285e-05,
+      "loss": 0.753,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921231977975958e-05,
+      "loss": 0.9262,
+      "step": 1331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9210866477899685e-05,
+      "loss": 1.0051,
+      "step": 1332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920941189164525e-05,
+      "loss": 0.7566,
+      "step": 1333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9207956021199113e-05,
+      "loss": 0.8219,
+      "step": 1334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920649886676429e-05,
+      "loss": 0.6106,
+      "step": 1335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9205040428543962e-05,
+      "loss": 0.704,
+      "step": 1336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.92035807067415e-05,
+      "loss": 1.0874,
+      "step": 1337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.920211970156046e-05,
+      "loss": 1.0247,
+      "step": 1338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9200657413204562e-05,
+      "loss": 0.9873,
+      "step": 1339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9199193841877713e-05,
+      "loss": 0.7408,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9197728987784003e-05,
+      "loss": 0.8395,
+      "step": 1341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9196262851127695e-05,
+      "loss": 0.7283,
+      "step": 1342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9194795432113225e-05,
+      "loss": 0.6873,
+      "step": 1343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9193326730945228e-05,
+      "loss": 0.9613,
+      "step": 1344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9191856747828494e-05,
+      "loss": 0.7636,
+      "step": 1345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9190385482968003e-05,
+      "loss": 0.8235,
+      "step": 1346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918891293656892e-05,
+      "loss": 0.6185,
+      "step": 1347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9187439108836573e-05,
+      "loss": 0.7455,
+      "step": 1348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9185963999976483e-05,
+      "loss": 1.0917,
+      "step": 1349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9184487610194343e-05,
+      "loss": 0.6898,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918300993969603e-05,
+      "loss": 0.881,
+      "step": 1351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918153098868759e-05,
+      "loss": 0.8243,
+      "step": 1352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9180050757375258e-05,
+      "loss": 0.8025,
+      "step": 1353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9178569245965434e-05,
+      "loss": 0.8111,
+      "step": 1354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917708645466471e-05,
+      "loss": 0.7485,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9175602383679855e-05,
+      "loss": 0.7007,
+      "step": 1356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9174117033217806e-05,
+      "loss": 0.5501,
+      "step": 1357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917263040348569e-05,
+      "loss": 0.825,
+      "step": 1358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9171142494690807e-05,
+      "loss": 0.8099,
+      "step": 1359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9169653307040636e-05,
+      "loss": 0.6978,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9168162840742833e-05,
+      "loss": 0.6442,
+      "step": 1361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9166671096005236e-05,
+      "loss": 1.1334,
+      "step": 1362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9165178073035857e-05,
+      "loss": 0.7394,
+      "step": 1363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916368377204289e-05,
+      "loss": 0.7812,
+      "step": 1364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.91621881932347e-05,
+      "loss": 1.0531,
+      "step": 1365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9160691336819842e-05,
+      "loss": 0.7547,
+      "step": 1366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915919320300704e-05,
+      "loss": 0.8614,
+      "step": 1367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91576937920052e-05,
+      "loss": 0.6787,
+      "step": 1368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9156193104023404e-05,
+      "loss": 0.6763,
+      "step": 1369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9154691139270912e-05,
+      "loss": 0.7763,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9153187897957167e-05,
+      "loss": 0.7581,
+      "step": 1371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9151683380291784e-05,
+      "loss": 0.6458,
+      "step": 1372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9150177586484554e-05,
+      "loss": 0.678,
+      "step": 1373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9148670516745456e-05,
+      "loss": 0.7237,
+      "step": 1374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9147162171284642e-05,
+      "loss": 0.6639,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914565255031243e-05,
+      "loss": 0.6345,
+      "step": 1376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914414165403934e-05,
+      "loss": 0.732,
+      "step": 1377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9142629482676048e-05,
+      "loss": 0.7911,
+      "step": 1378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9141116036433417e-05,
+      "loss": 0.6575,
+      "step": 1379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9139601315522492e-05,
+      "loss": 0.7907,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9138085320154488e-05,
+      "loss": 0.8285,
+      "step": 1381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9136568050540797e-05,
+      "loss": 0.7919,
+      "step": 1382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9135049506893e-05,
+      "loss": 0.7982,
+      "step": 1383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913352968942284e-05,
+      "loss": 0.9121,
+      "step": 1384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9132008598342252e-05,
+      "loss": 0.7168,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9130486233863338e-05,
+      "loss": 0.8351,
+      "step": 1386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912896259619838e-05,
+      "loss": 1.1451,
+      "step": 1387
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912743768555984e-05,
+      "loss": 0.9182,
+      "step": 1388
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9125911502160365e-05,
+      "loss": 0.7482,
+      "step": 1389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9124384046212765e-05,
+      "loss": 0.6077,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912285531793003e-05,
+      "loss": 0.8117,
+      "step": 1391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9121325317525335e-05,
+      "loss": 0.7787,
+      "step": 1392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911979404521203e-05,
+      "loss": 0.8946,
+      "step": 1393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9118261501203638e-05,
+      "loss": 1.1281,
+      "step": 1394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9116727685713862e-05,
+      "loss": 0.627,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911519259895658e-05,
+      "loss": 1.0931,
+      "step": 1396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9113656241145858e-05,
+      "loss": 0.7443,
+      "step": 1397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9112118612495926e-05,
+      "loss": 0.7676,
+      "step": 1398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91105797132212e-05,
+      "loss": 0.7358,
+      "step": 1399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9109039543536258e-05,
+      "loss": 0.6622,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9107498103655878e-05,
+      "loss": 1.0281,
+      "step": 1401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9105955393795e-05,
+      "loss": 0.6196,
+      "step": 1402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9104411414168743e-05,
+      "loss": 0.7531,
+      "step": 1403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9102866164992406e-05,
+      "loss": 1.041,
+      "step": 1404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9101319646481466e-05,
+      "loss": 0.7841,
+      "step": 1405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.909977185885157e-05,
+      "loss": 0.621,
+      "step": 1406
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9098222802318552e-05,
+      "loss": 1.0925,
+      "step": 1407
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9096672477098416e-05,
+      "loss": 0.7298,
+      "step": 1408
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.909512088340734e-05,
+      "loss": 0.6296,
+      "step": 1409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9093568021461693e-05,
+      "loss": 0.8467,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9092013891478004e-05,
+      "loss": 0.8014,
+      "step": 1411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.909045849367299e-05,
+      "loss": 0.8545,
+      "step": 1412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9088901828263533e-05,
+      "loss": 0.7251,
+      "step": 1413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9087343895466705e-05,
+      "loss": 0.6824,
+      "step": 1414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9085784695499755e-05,
+      "loss": 0.6237,
+      "step": 1415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90842242285801e-05,
+      "loss": 0.6914,
+      "step": 1416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908266249492533e-05,
+      "loss": 0.7637,
+      "step": 1417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9081099494753225e-05,
+      "loss": 0.6053,
+      "step": 1418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9079535228281734e-05,
+      "loss": 0.6491,
+      "step": 1419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9077969695728985e-05,
+      "loss": 0.825,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907640289731328e-05,
+      "loss": 1.0599,
+      "step": 1421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9074834833253092e-05,
+      "loss": 0.7595,
+      "step": 1422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9073265503767086e-05,
+      "loss": 0.6562,
+      "step": 1423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907169490907409e-05,
+      "loss": 0.8204,
+      "step": 1424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9070123049393122e-05,
+      "loss": 0.8369,
+      "step": 1425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9068549924943352e-05,
+      "loss": 0.7609,
+      "step": 1426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9066975535944157e-05,
+      "loss": 0.8043,
+      "step": 1427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9065399882615066e-05,
+      "loss": 0.6381,
+      "step": 1428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9063822965175794e-05,
+      "loss": 0.8093,
+      "step": 1429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9062244783846236e-05,
+      "loss": 1.1671,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9060665338846457e-05,
+      "loss": 0.6992,
+      "step": 1431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9059084630396697e-05,
+      "loss": 0.7275,
+      "step": 1432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9057502658717378e-05,
+      "loss": 0.7248,
+      "step": 1433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9055919424029098e-05,
+      "loss": 1.1134,
+      "step": 1434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9054334926552625e-05,
+      "loss": 0.7577,
+      "step": 1435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905274916650891e-05,
+      "loss": 0.7256,
+      "step": 1436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9051162144119072e-05,
+      "loss": 1.1093,
+      "step": 1437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9049573859604415e-05,
+      "loss": 0.9753,
+      "step": 1438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9047984313186408e-05,
+      "loss": 0.6458,
+      "step": 1439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9046393505086714e-05,
+      "loss": 1.0466,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9044801435527157e-05,
+      "loss": 0.9111,
+      "step": 1441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904320810472973e-05,
+      "loss": 0.8153,
+      "step": 1442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9041613512916627e-05,
+      "loss": 0.6737,
+      "step": 1443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9040017660310196e-05,
+      "loss": 0.7633,
+      "step": 1444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903842054713297e-05,
+      "loss": 0.6755,
+      "step": 1445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9036822173607653e-05,
+      "loss": 0.8482,
+      "step": 1446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903522253995713e-05,
+      "loss": 0.7693,
+      "step": 1447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903362164640446e-05,
+      "loss": 0.8702,
+      "step": 1448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9032019493172877e-05,
+      "loss": 0.5496,
+      "step": 1449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9030416080485786e-05,
+      "loss": 0.876,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.902881140856678e-05,
+      "loss": 0.6769,
+      "step": 1451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9027205477639616e-05,
+      "loss": 0.7494,
+      "step": 1452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902559828792823e-05,
+      "loss": 0.7507,
+      "step": 1453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9023989839656733e-05,
+      "loss": 0.7557,
+      "step": 1454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9022380133049416e-05,
+      "loss": 0.726,
+      "step": 1455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9020769168330737e-05,
+      "loss": 0.6492,
+      "step": 1456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9019156945725342e-05,
+      "loss": 0.7206,
+      "step": 1457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9017543465458038e-05,
+      "loss": 1.0604,
+      "step": 1458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9015928727753814e-05,
+      "loss": 1.0895,
+      "step": 1459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9014312732837837e-05,
+      "loss": 0.7279,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901269548093545e-05,
+      "loss": 0.6689,
+      "step": 1461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9011076972272162e-05,
+      "loss": 1.1519,
+      "step": 1462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900945720707367e-05,
+      "loss": 1.0678,
+      "step": 1463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900783618556583e-05,
+      "loss": 0.803,
+      "step": 1464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9006213907974694e-05,
+      "loss": 0.8922,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9004590374526467e-05,
+      "loss": 0.6371,
+      "step": 1466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9002965585447552e-05,
+      "loss": 1.0758,
+      "step": 1467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9001339540964504e-05,
+      "loss": 0.6694,
+      "step": 1468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8999712241304067e-05,
+      "loss": 1.2099,
+      "step": 1469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899808368669316e-05,
+      "loss": 1.1116,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8996453877358876e-05,
+      "loss": 0.6813,
+      "step": 1471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8994822813528478e-05,
+      "loss": 0.8455,
+      "step": 1472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8993190495429404e-05,
+      "loss": 0.7227,
+      "step": 1473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8991556923289276e-05,
+      "loss": 1.0807,
+      "step": 1474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898992209733588e-05,
+      "loss": 0.643,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8988286017797186e-05,
+      "loss": 0.607,
+      "step": 1476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898664868490133e-05,
+      "loss": 0.7232,
+      "step": 1477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8985010098876632e-05,
+      "loss": 0.6598,
+      "step": 1478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8983370259951575e-05,
+      "loss": 0.6489,
+      "step": 1479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8981729168354833e-05,
+      "loss": 0.7431,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898008682431524e-05,
+      "loss": 0.922,
+      "step": 1481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897844322806181e-05,
+      "loss": 0.7239,
+      "step": 1482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897679837982373e-05,
+      "loss": 0.8181,
+      "step": 1483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897515227983037e-05,
+      "loss": 0.8615,
+      "step": 1484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8973504928311262e-05,
+      "loss": 0.7101,
+      "step": 1485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897185632549612e-05,
+      "loss": 1.2012,
+      "step": 1486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897020647161483e-05,
+      "loss": 1.1325,
+      "step": 1487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896855536689745e-05,
+      "loss": 0.7696,
+      "step": 1488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8966903011574225e-05,
+      "loss": 0.9228,
+      "step": 1489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896524940587556e-05,
+      "loss": 0.7738,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896359455003203e-05,
+      "loss": 0.9287,
+      "step": 1491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896193844427441e-05,
+      "loss": 0.7427,
+      "step": 1492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8960281088833624e-05,
+      "loss": 1.0931,
+      "step": 1493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895862248394078e-05,
+      "loss": 0.9955,
+      "step": 1494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.895696262982716e-05,
+      "loss": 0.7012,
+      "step": 1495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8955301526724224e-05,
+      "loss": 0.7546,
+      "step": 1496
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8953639174863597e-05,
+      "loss": 0.9531,
+      "step": 1497
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895197557447708e-05,
+      "loss": 0.7869,
+      "step": 1498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8950310725796655e-05,
+      "loss": 0.638,
+      "step": 1499
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8948644629054477e-05,
+      "loss": 0.6055,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894697728448287e-05,
+      "loss": 0.8757,
+      "step": 1501
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894530869231433e-05,
+      "loss": 0.9991,
+      "step": 1502
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894363885278154e-05,
+      "loss": 0.7323,
+      "step": 1503
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894196776611734e-05,
+      "loss": 0.6333,
+      "step": 1504
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.894029543255475e-05,
+      "loss": 0.7209,
+      "step": 1505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8938621852326974e-05,
+      "loss": 0.6315,
+      "step": 1506
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8936947025667382e-05,
+      "loss": 0.8056,
+      "step": 1507
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893527095280951e-05,
+      "loss": 0.745,
+      "step": 1508
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893359363398708e-05,
+      "loss": 0.9115,
+      "step": 1509
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8931915069433983e-05,
+      "loss": 0.7156,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.893023525938428e-05,
+      "loss": 1.032,
+      "step": 1511
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8928554204072217e-05,
+      "loss": 0.7454,
+      "step": 1512
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.89268719037322e-05,
+      "loss": 0.9598,
+      "step": 1513
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8925188358598815e-05,
+      "loss": 1.0997,
+      "step": 1514
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8923503568906824e-05,
+      "loss": 0.6372,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8921817534891158e-05,
+      "loss": 0.6175,
+      "step": 1516
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892013025678692e-05,
+      "loss": 0.7323,
+      "step": 1517
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8918441734829397e-05,
+      "loss": 0.7013,
+      "step": 1518
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8916751969254036e-05,
+      "loss": 0.8159,
+      "step": 1519
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8915060960296467e-05,
+      "loss": 0.9906,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8913368708192494e-05,
+      "loss": 0.844,
+      "step": 1521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891167521317808e-05,
+      "loss": 0.8214,
+      "step": 1522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890998047548938e-05,
+      "loss": 0.6562,
+      "step": 1523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890828449536271e-05,
+      "loss": 0.6676,
+      "step": 1524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8906587273034566e-05,
+      "loss": 1.0782,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890488880874161e-05,
+      "loss": 0.6907,
+      "step": 1526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8903189102720687e-05,
+      "loss": 0.8433,
+      "step": 1527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890148815520881e-05,
+      "loss": 0.8792,
+      "step": 1528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8899785966443157e-05,
+      "loss": 0.7869,
+      "step": 1529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8898082536661097e-05,
+      "loss": 0.7778,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889637786610015e-05,
+      "loss": 0.8278,
+      "step": 1531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8894671954998038e-05,
+      "loss": 0.8032,
+      "step": 1532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8892964803592624e-05,
+      "loss": 0.7085,
+      "step": 1533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8891256412121966e-05,
+      "loss": 0.7316,
+      "step": 1534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8889546780824287e-05,
+      "loss": 0.627,
+      "step": 1535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8887835909937983e-05,
+      "loss": 0.9437,
+      "step": 1536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8886123799701622e-05,
+      "loss": 1.1596,
+      "step": 1537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888441045035395e-05,
+      "loss": 0.8312,
+      "step": 1538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888269586213388e-05,
+      "loss": 1.3138,
+      "step": 1539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8880980035280497e-05,
+      "loss": 0.688,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887926297003307e-05,
+      "loss": 0.9083,
+      "step": 1541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8877544666631027e-05,
+      "loss": 0.7641,
+      "step": 1542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8875825125313976e-05,
+      "loss": 0.7615,
+      "step": 1543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8874104346321693e-05,
+      "loss": 0.7751,
+      "step": 1544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887238232989413e-05,
+      "loss": 0.7192,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8870659076271414e-05,
+      "loss": 0.8313,
+      "step": 1546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8868934585693838e-05,
+      "loss": 0.6826,
+      "step": 1547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8867208858401867e-05,
+      "loss": 1.081,
+      "step": 1548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8865481894636154e-05,
+      "loss": 1.1848,
+      "step": 1549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8863753694637502e-05,
+      "loss": 0.6564,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8862024258646903e-05,
+      "loss": 1.1259,
+      "step": 1551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8860293586905508e-05,
+      "loss": 0.897,
+      "step": 1552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8858561679654662e-05,
+      "loss": 0.7423,
+      "step": 1553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8856828537135852e-05,
+      "loss": 0.8429,
+      "step": 1554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885509415959076e-05,
+      "loss": 0.6255,
+      "step": 1555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8853358547261238e-05,
+      "loss": 1.0269,
+      "step": 1556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88516217003893e-05,
+      "loss": 1.0699,
+      "step": 1557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8849883619217138e-05,
+      "loss": 0.6928,
+      "step": 1558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8848144303987118e-05,
+      "loss": 1.0098,
+      "step": 1559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8846403754941778e-05,
+      "loss": 0.7539,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.884466197232382e-05,
+      "loss": 0.8034,
+      "step": 1561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8842918956376127e-05,
+      "loss": 0.9888,
+      "step": 1562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8841174707341755e-05,
+      "loss": 0.6651,
+      "step": 1563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883942922546392e-05,
+      "loss": 0.859,
+      "step": 1564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8837682510986027e-05,
+      "loss": 0.7757,
+      "step": 1565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8835934564151637e-05,
+      "loss": 0.7114,
+      "step": 1566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8834185385204495e-05,
+      "loss": 0.724,
+      "step": 1567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8832434974388504e-05,
+      "loss": 0.7959,
+      "step": 1568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8830683331947756e-05,
+      "loss": 0.9846,
+      "step": 1569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88289304581265e-05,
+      "loss": 0.8614,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882717635316917e-05,
+      "loss": 1.0993,
+      "step": 1571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882542101732036e-05,
+      "loss": 0.6669,
+      "step": 1572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882366445082483e-05,
+      "loss": 0.6545,
+      "step": 1573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882190665392754e-05,
+      "loss": 1.0927,
+      "step": 1574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8820147626873595e-05,
+      "loss": 0.7816,
+      "step": 1575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8818387369908274e-05,
+      "loss": 0.7147,
+      "step": 1576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8816625883277044e-05,
+      "loss": 0.7472,
+      "step": 1577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8814863167225524e-05,
+      "loss": 0.9857,
+      "step": 1578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8813099221999514e-05,
+      "loss": 0.9681,
+      "step": 1579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8811334047844992e-05,
+      "loss": 0.5677,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8809567645008088e-05,
+      "loss": 1.1218,
+      "step": 1581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8807800013735127e-05,
+      "loss": 0.7339,
+      "step": 1582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8806031154272585e-05,
+      "loss": 0.9451,
+      "step": 1583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8804261066867124e-05,
+      "loss": 0.7265,
+      "step": 1584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8802489751765568e-05,
+      "loss": 0.7062,
+      "step": 1585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8800717209214913e-05,
+      "loss": 0.938,
+      "step": 1586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879894343946233e-05,
+      "loss": 1.1814,
+      "step": 1587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8797168442755166e-05,
+      "loss": 0.9354,
+      "step": 1588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879539221934092e-05,
+      "loss": 0.8612,
+      "step": 1589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8793614769467285e-05,
+      "loss": 0.7867,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879183609338211e-05,
+      "loss": 0.6585,
+      "step": 1591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8790056191333417e-05,
+      "loss": 0.8589,
+      "step": 1592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8788275063569412e-05,
+      "loss": 0.9218,
+      "step": 1593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878649271033845e-05,
+      "loss": 0.678,
+      "step": 1594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8784709131889076e-05,
+      "loss": 0.6409,
+      "step": 1595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8782924328469994e-05,
+      "loss": 0.8981,
+      "step": 1596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8781138300330086e-05,
+      "loss": 1.0046,
+      "step": 1597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87793510477184e-05,
+      "loss": 0.9692,
+      "step": 1598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8777562570884154e-05,
+      "loss": 0.9844,
+      "step": 1599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8775772870076744e-05,
+      "loss": 0.8953,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877398194554573e-05,
+      "loss": 0.7177,
+      "step": 1601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8772189797540842e-05,
+      "loss": 0.8146,
+      "step": 1602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877039642631199e-05,
+      "loss": 1.0647,
+      "step": 1603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876860183210924e-05,
+      "loss": 0.7042,
+      "step": 1604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8766806015182846e-05,
+      "loss": 0.6664,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876500897578321e-05,
+      "loss": 0.7086,
+      "step": 1606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876321071416093e-05,
+      "loss": 0.6928,
+      "step": 1607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8761411230566752e-05,
+      "loss": 0.7552,
+      "step": 1608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8759610525251605e-05,
+      "loss": 0.881,
+      "step": 1609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8757808598466593e-05,
+      "loss": 0.6652,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8756005450462972e-05,
+      "loss": 0.874,
+      "step": 1611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875420108149218e-05,
+      "loss": 0.6758,
+      "step": 1612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.875239549180583e-05,
+      "loss": 0.6111,
+      "step": 1613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.87505886816557e-05,
+      "loss": 0.6462,
+      "step": 1614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874878065129373e-05,
+      "loss": 0.7028,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8746971400972047e-05,
+      "loss": 0.7644,
+      "step": 1616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8745160930942933e-05,
+      "loss": 1.0125,
+      "step": 1617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874334924145885e-05,
+      "loss": 0.7994,
+      "step": 1618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.874153633277242e-05,
+      "loss": 0.7671,
+      "step": 1619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873972220513645e-05,
+      "loss": 0.7044,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8737906858803902e-05,
+      "loss": 1.051,
+      "step": 1621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8736090294027912e-05,
+      "loss": 0.5822,
+      "step": 1622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8734272511061796e-05,
+      "loss": 0.9544,
+      "step": 1623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8732453510159025e-05,
+      "loss": 0.75,
+      "step": 1624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8730633291573247e-05,
+      "loss": 0.7162,
+      "step": 1625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8728811855558286e-05,
+      "loss": 0.834,
+      "step": 1626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8726989202368123e-05,
+      "loss": 0.7994,
+      "step": 1627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8725165332256913e-05,
+      "loss": 0.8063,
+      "step": 1628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872334024547899e-05,
+      "loss": 0.7351,
+      "step": 1629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872151394228885e-05,
+      "loss": 0.6418,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8719686422941147e-05,
+      "loss": 0.6949,
+      "step": 1631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871785768769073e-05,
+      "loss": 0.7465,
+      "step": 1632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8716027736792597e-05,
+      "loss": 0.9427,
+      "step": 1633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8714196570501925e-05,
+      "loss": 0.7598,
+      "step": 1634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.871236418907406e-05,
+      "loss": 0.9823,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8710530592764514e-05,
+      "loss": 0.6979,
+      "step": 1636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8708695781828967e-05,
+      "loss": 0.949,
+      "step": 1637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870685975652327e-05,
+      "loss": 0.6917,
+      "step": 1638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870502251710345e-05,
+      "loss": 0.7882,
+      "step": 1639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8703184063825698e-05,
+      "loss": 0.9441,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8701344396946372e-05,
+      "loss": 0.7701,
+      "step": 1641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8699503516721998e-05,
+      "loss": 0.7264,
+      "step": 1642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8697661423409282e-05,
+      "loss": 0.9529,
+      "step": 1643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8695818117265085e-05,
+      "loss": 0.7945,
+      "step": 1644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8693973598546447e-05,
+      "loss": 0.7477,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8692127867510576e-05,
+      "loss": 0.8128,
+      "step": 1646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8690280924414843e-05,
+      "loss": 0.7647,
+      "step": 1647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8688432769516795e-05,
+      "loss": 1.0956,
+      "step": 1648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868658340307414e-05,
+      "loss": 0.6533,
+      "step": 1649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8684732825344765e-05,
+      "loss": 0.8395,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8682881036586722e-05,
+      "loss": 0.8464,
+      "step": 1651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8681028037058226e-05,
+      "loss": 0.6431,
+      "step": 1652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867917382701767e-05,
+      "loss": 0.6774,
+      "step": 1653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8677318406723606e-05,
+      "loss": 0.9111,
+      "step": 1654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8675461776434765e-05,
+      "loss": 0.6736,
+      "step": 1655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.867360393641004e-05,
+      "loss": 0.7544,
+      "step": 1656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8671744886908494e-05,
+      "loss": 0.8243,
+      "step": 1657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866988462818936e-05,
+      "loss": 1.146,
+      "step": 1658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866802316051204e-05,
+      "loss": 0.6129,
+      "step": 1659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.86661604841361e-05,
+      "loss": 0.8894,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8664296599321278e-05,
+      "loss": 1.0319,
+      "step": 1661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8662431506327484e-05,
+      "loss": 0.9174,
+      "step": 1662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8660565205414793e-05,
+      "loss": 0.7582,
+      "step": 1663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8658697696843445e-05,
+      "loss": 0.8424,
+      "step": 1664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8656828980873854e-05,
+      "loss": 0.9322,
+      "step": 1665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8654959057766598e-05,
+      "loss": 0.7054,
+      "step": 1666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8653087927782428e-05,
+      "loss": 0.7865,
+      "step": 1667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8651215591182258e-05,
+      "loss": 0.7508,
+      "step": 1668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8649342048227175e-05,
+      "loss": 1.0085,
+      "step": 1669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8647467299178434e-05,
+      "loss": 0.8306,
+      "step": 1670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864559134429745e-05,
+      "loss": 0.6005,
+      "step": 1671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8643714183845822e-05,
+      "loss": 0.6195,
+      "step": 1672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.86418358180853e-05,
+      "loss": 0.8151,
+      "step": 1673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8639956247277812e-05,
+      "loss": 0.6152,
+      "step": 1674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863807547168545e-05,
+      "loss": 0.7302,
+      "step": 1675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863619349157048e-05,
+      "loss": 1.0407,
+      "step": 1676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8634310307195326e-05,
+      "loss": 1.0866,
+      "step": 1677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863242591882259e-05,
+      "loss": 0.6079,
+      "step": 1678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8630540326715038e-05,
+      "loss": 0.9426,
+      "step": 1679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8628653531135602e-05,
+      "loss": 0.8276,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862676553234738e-05,
+      "loss": 0.7554,
+      "step": 1681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8624876330613645e-05,
+      "loss": 0.5311,
+      "step": 1682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8622985926197832e-05,
+      "loss": 0.7188,
+      "step": 1683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862109431936355e-05,
+      "loss": 0.6999,
+      "step": 1684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8619201510374565e-05,
+      "loss": 1.0444,
+      "step": 1685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8617307499494815e-05,
+      "loss": 0.9297,
+      "step": 1686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8615412286988418e-05,
+      "loss": 0.6123,
+      "step": 1687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8613515873119636e-05,
+      "loss": 0.7692,
+      "step": 1688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861161825815292e-05,
+      "loss": 0.7063,
+      "step": 1689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8609719442352875e-05,
+      "loss": 0.7698,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860781942598428e-05,
+      "loss": 0.807,
+      "step": 1691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8605918209312084e-05,
+      "loss": 0.6669,
+      "step": 1692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8604015792601395e-05,
+      "loss": 0.6843,
+      "step": 1693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860211217611749e-05,
+      "loss": 0.7658,
+      "step": 1694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8600207360125823e-05,
+      "loss": 0.6045,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8598301344892e-05,
+      "loss": 0.7747,
+      "step": 1696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859639413068181e-05,
+      "loss": 0.9124,
+      "step": 1697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8594485717761193e-05,
+      "loss": 0.7117,
+      "step": 1698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859257610639627e-05,
+      "loss": 0.7071,
+      "step": 1699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8590665296853324e-05,
+      "loss": 0.6221,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8588753289398804e-05,
+      "loss": 0.7414,
+      "step": 1701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8586840084299323e-05,
+      "loss": 1.1946,
+      "step": 1702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8584925681821672e-05,
+      "loss": 0.7656,
+      "step": 1703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8583010082232797e-05,
+      "loss": 0.7616,
+      "step": 1704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8581093285799814e-05,
+      "loss": 0.7694,
+      "step": 1705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857917529279001e-05,
+      "loss": 0.8122,
+      "step": 1706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857725610347084e-05,
+      "loss": 0.6479,
+      "step": 1707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857533571810992e-05,
+      "loss": 0.6844,
+      "step": 1708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8573414136975025e-05,
+      "loss": 0.7643,
+      "step": 1709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8571491360334123e-05,
+      "loss": 0.758,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8569567388455326e-05,
+      "loss": 0.7482,
+      "step": 1711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567642221606916e-05,
+      "loss": 0.7608,
+      "step": 1712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8565715860057348e-05,
+      "loss": 1.1189,
+      "step": 1713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.856378830407524e-05,
+      "loss": 0.6954,
+      "step": 1714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8561859553929377e-05,
+      "loss": 0.6899,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855992960988871e-05,
+      "loss": 0.702,
+      "step": 1716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8557998472222358e-05,
+      "loss": 0.712,
+      "step": 1717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85560661411996e-05,
+      "loss": 0.6742,
+      "step": 1718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8554132617089896e-05,
+      "loss": 0.8809,
+      "step": 1719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8552197900162855e-05,
+      "loss": 0.7102,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.855026199068827e-05,
+      "loss": 0.6282,
+      "step": 1721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.854832488893608e-05,
+      "loss": 0.9118,
+      "step": 1722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8546386595176405e-05,
+      "loss": 0.9022,
+      "step": 1723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8544447109679532e-05,
+      "loss": 0.7765,
+      "step": 1724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8542506432715902e-05,
+      "loss": 0.6809,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8540564564556134e-05,
+      "loss": 0.6805,
+      "step": 1726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8538621505471007e-05,
+      "loss": 0.6324,
+      "step": 1727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8536677255731473e-05,
+      "loss": 1.0354,
+      "step": 1728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8534731815608637e-05,
+      "loss": 0.7667,
+      "step": 1729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853278518537378e-05,
+      "loss": 0.919,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853083736529835e-05,
+      "loss": 0.6344,
+      "step": 1731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8528888355653954e-05,
+      "loss": 0.7338,
+      "step": 1732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8526938156712372e-05,
+      "loss": 0.7565,
+      "step": 1733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8524986768745546e-05,
+      "loss": 0.6889,
+      "step": 1734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852303419202558e-05,
+      "loss": 0.8664,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852108042682475e-05,
+      "loss": 0.6436,
+      "step": 1736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8519125473415498e-05,
+      "loss": 0.7728,
+      "step": 1737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8517169332070425e-05,
+      "loss": 0.756,
+      "step": 1738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8515212003062307e-05,
+      "loss": 0.7477,
+      "step": 1739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851325348666408e-05,
+      "loss": 0.7146,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8511293783148842e-05,
+      "loss": 0.8554,
+      "step": 1741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8509332892789864e-05,
+      "loss": 0.7772,
+      "step": 1742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8507370815860576e-05,
+      "loss": 0.8113,
+      "step": 1743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850540755263458e-05,
+      "loss": 0.7712,
+      "step": 1744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850344310338564e-05,
+      "loss": 0.7642,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8501477468387684e-05,
+      "loss": 0.5874,
+      "step": 1746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8499510647914812e-05,
+      "loss": 0.6179,
+      "step": 1747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8497542642241274e-05,
+      "loss": 0.9943,
+      "step": 1748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.84955734516415e-05,
+      "loss": 0.7366,
+      "step": 1749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8493603076390087e-05,
+      "loss": 0.7307,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8491631516761783e-05,
+      "loss": 0.8474,
+      "step": 1751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.848965877303151e-05,
+      "loss": 0.6683,
+      "step": 1752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8487684845474358e-05,
+      "loss": 0.8276,
+      "step": 1753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8485709734365574e-05,
+      "loss": 1.0613,
+      "step": 1754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8483733439980572e-05,
+      "loss": 0.7321,
+      "step": 1755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8481755962594943e-05,
+      "loss": 0.7629,
+      "step": 1756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8479777302484426e-05,
+      "loss": 0.685,
+      "step": 1757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8477797459924934e-05,
+      "loss": 0.7819,
+      "step": 1758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847581643519254e-05,
+      "loss": 0.7365,
+      "step": 1759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8473834228563488e-05,
+      "loss": 0.7346,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8471850840314182e-05,
+      "loss": 0.7593,
+      "step": 1761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8469866270721194e-05,
+      "loss": 0.8475,
+      "step": 1762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846788052006126e-05,
+      "loss": 1.0509,
+      "step": 1763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8465893588611275e-05,
+      "loss": 0.8627,
+      "step": 1764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846390547664831e-05,
+      "loss": 0.7223,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846191618444959e-05,
+      "loss": 0.6132,
+      "step": 1766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8459925712292507e-05,
+      "loss": 0.6501,
+      "step": 1767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8457934060454623e-05,
+      "loss": 0.9941,
+      "step": 1768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8455941229213657e-05,
+      "loss": 0.7588,
+      "step": 1769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.84539472188475e-05,
+      "loss": 0.7579,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8451952029634204e-05,
+      "loss": 0.7966,
+      "step": 1771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8449955661851983e-05,
+      "loss": 0.7805,
+      "step": 1772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8447958115779214e-05,
+      "loss": 0.7212,
+      "step": 1773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.844595939169445e-05,
+      "loss": 0.6072,
+      "step": 1774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.844395948987639e-05,
+      "loss": 0.681,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8441958410603913e-05,
+      "loss": 0.6947,
+      "step": 1776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8439956154156055e-05,
+      "loss": 0.7179,
+      "step": 1777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8437952720812018e-05,
+      "loss": 0.7692,
+      "step": 1778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.843594811085117e-05,
+      "loss": 0.9979,
+      "step": 1779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8433942324553035e-05,
+      "loss": 0.664,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8431935362197307e-05,
+      "loss": 0.7019,
+      "step": 1781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8429927224063852e-05,
+      "loss": 0.799,
+      "step": 1782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842791791043268e-05,
+      "loss": 0.7562,
+      "step": 1783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8425907421583987e-05,
+      "loss": 0.6511,
+      "step": 1784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8423895757798112e-05,
+      "loss": 0.7874,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842188291935558e-05,
+      "loss": 1.0441,
+      "step": 1786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8419868906537058e-05,
+      "loss": 0.9125,
+      "step": 1787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8417853719623397e-05,
+      "loss": 0.7624,
+      "step": 1788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8415837358895587e-05,
+      "loss": 0.6706,
+      "step": 1789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841381982463481e-05,
+      "loss": 0.9278,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841180111712239e-05,
+      "loss": 1.0997,
+      "step": 1791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8409781236639825e-05,
+      "loss": 1.0507,
+      "step": 1792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8407760183468774e-05,
+      "loss": 1.0238,
+      "step": 1793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8405737957891058e-05,
+      "loss": 0.658,
+      "step": 1794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8403714560188665e-05,
+      "loss": 1.0861,
+      "step": 1795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8401689990643745e-05,
+      "loss": 0.8435,
+      "step": 1796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8399664249538608e-05,
+      "loss": 0.7587,
+      "step": 1797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8397637337155735e-05,
+      "loss": 1.0384,
+      "step": 1798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8395609253777758e-05,
+      "loss": 0.7782,
+      "step": 1799
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8393579999687484e-05,
+      "loss": 0.958,
+      "step": 1800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8391549575167883e-05,
+      "loss": 0.804,
+      "step": 1801
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8389517980502077e-05,
+      "loss": 1.1422,
+      "step": 1802
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8387485215973363e-05,
+      "loss": 0.6708,
+      "step": 1803
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83854512818652e-05,
+      "loss": 0.741,
+      "step": 1804
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83834161784612e-05,
+      "loss": 0.598,
+      "step": 1805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8381379906045142e-05,
+      "loss": 1.0435,
+      "step": 1806
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8379342464900982e-05,
+      "loss": 0.6659,
+      "step": 1807
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837730385531282e-05,
+      "loss": 0.9554,
+      "step": 1808
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8375264077564926e-05,
+      "loss": 0.9033,
+      "step": 1809
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8373223131941738e-05,
+      "loss": 0.8339,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8371181018727847e-05,
+      "loss": 1.071,
+      "step": 1811
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.836913773820802e-05,
+      "loss": 0.7682,
+      "step": 1812
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8367093290667172e-05,
+      "loss": 0.7287,
+      "step": 1813
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8365047676390394e-05,
+      "loss": 0.7781,
+      "step": 1814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8363000895662925e-05,
+      "loss": 0.6887,
+      "step": 1815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8360952948770177e-05,
+      "loss": 0.7218,
+      "step": 1816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835890383599773e-05,
+      "loss": 0.8209,
+      "step": 1817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8356853557631312e-05,
+      "loss": 0.8397,
+      "step": 1818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8354802113956826e-05,
+      "loss": 1.059,
+      "step": 1819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8352749505260325e-05,
+      "loss": 0.7961,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.835069573182804e-05,
+      "loss": 0.7462,
+      "step": 1821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8348640793946352e-05,
+      "loss": 1.0079,
+      "step": 1822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.834658469190181e-05,
+      "loss": 0.7183,
+      "step": 1823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8344527425981117e-05,
+      "loss": 0.6965,
+      "step": 1824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8342468996471154e-05,
+      "loss": 0.7363,
+      "step": 1825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8340409403658954e-05,
+      "loss": 0.7535,
+      "step": 1826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833834864783171e-05,
+      "loss": 0.6547,
+      "step": 1827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8336286729276788e-05,
+      "loss": 0.9074,
+      "step": 1828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83342236482817e-05,
+      "loss": 1.0287,
+      "step": 1829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833215940513413e-05,
+      "loss": 0.8538,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8330094000121925e-05,
+      "loss": 0.8347,
+      "step": 1831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8328027433533098e-05,
+      "loss": 0.9671,
+      "step": 1832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8325959705655808e-05,
+      "loss": 0.6458,
+      "step": 1833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83238908167784e-05,
+      "loss": 1.156,
+      "step": 1834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8321820767189348e-05,
+      "loss": 0.9201,
+      "step": 1835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831974955717732e-05,
+      "loss": 0.9159,
+      "step": 1836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8317677187031132e-05,
+      "loss": 0.6851,
+      "step": 1837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8315603657039757e-05,
+      "loss": 0.8063,
+      "step": 1838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8313528967492335e-05,
+      "loss": 0.7478,
+      "step": 1839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8311453118678176e-05,
+      "loss": 1.0574,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8309376110886735e-05,
+      "loss": 0.7237,
+      "step": 1841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8307297944407637e-05,
+      "loss": 1.0769,
+      "step": 1842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8305218619530677e-05,
+      "loss": 0.7357,
+      "step": 1843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8303138136545792e-05,
+      "loss": 0.8117,
+      "step": 1844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.83010564957431e-05,
+      "loss": 0.6598,
+      "step": 1845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829897369741287e-05,
+      "loss": 0.6677,
+      "step": 1846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829688974184553e-05,
+      "loss": 0.7042,
+      "step": 1847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829480462933168e-05,
+      "loss": 0.6929,
+      "step": 1848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8292718360162076e-05,
+      "loss": 0.6607,
+      "step": 1849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829063093462763e-05,
+      "loss": 0.7439,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828854235301942e-05,
+      "loss": 1.1084,
+      "step": 1851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8286452615628683e-05,
+      "loss": 0.7685,
+      "step": 1852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8284361722746824e-05,
+      "loss": 0.6697,
+      "step": 1853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8282269674665403e-05,
+      "loss": 0.9202,
+      "step": 1854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8280176471676137e-05,
+      "loss": 0.7076,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.827808211407092e-05,
+      "loss": 0.8164,
+      "step": 1856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8275986602141788e-05,
+      "loss": 0.8479,
+      "step": 1857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8273889936180947e-05,
+      "loss": 0.6915,
+      "step": 1858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8271792116480767e-05,
+      "loss": 0.581,
+      "step": 1859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826969314333377e-05,
+      "loss": 0.7125,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826759301703265e-05,
+      "loss": 1.084,
+      "step": 1861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826549173787025e-05,
+      "loss": 0.7689,
+      "step": 1862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8263389306139583e-05,
+      "loss": 0.7586,
+      "step": 1863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8261285722133816e-05,
+      "loss": 1.1539,
+      "step": 1864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8259180986146282e-05,
+      "loss": 0.6391,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8257075098470477e-05,
+      "loss": 0.8614,
+      "step": 1866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8254968059400045e-05,
+      "loss": 0.635,
+      "step": 1867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8252859869228806e-05,
+      "loss": 0.665,
+      "step": 1868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8250750528250723e-05,
+      "loss": 0.6492,
+      "step": 1869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8248640036759942e-05,
+      "loss": 1.0377,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.824652839505075e-05,
+      "loss": 0.8007,
+      "step": 1871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8244415603417603e-05,
+      "loss": 0.657,
+      "step": 1872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8242301662155117e-05,
+      "loss": 0.8115,
+      "step": 1873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8240186571558063e-05,
+      "loss": 0.7596,
+      "step": 1874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8238070331921385e-05,
+      "loss": 0.9212,
+      "step": 1875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823595294354017e-05,
+      "loss": 1.2347,
+      "step": 1876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8233834406709678e-05,
+      "loss": 0.8583,
+      "step": 1877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823171472172532e-05,
+      "loss": 0.7154,
+      "step": 1878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8229593888882677e-05,
+      "loss": 0.7944,
+      "step": 1879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8227471908477485e-05,
+      "loss": 0.8591,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822534878080564e-05,
+      "loss": 0.623,
+      "step": 1881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8223224506163197e-05,
+      "loss": 0.6684,
+      "step": 1882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8221099084846374e-05,
+      "loss": 0.7271,
+      "step": 1883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821897251715154e-05,
+      "loss": 0.7794,
+      "step": 1884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821684480337524e-05,
+      "loss": 0.7107,
+      "step": 1885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821471594381417e-05,
+      "loss": 1.1791,
+      "step": 1886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8212585938765177e-05,
+      "loss": 0.6639,
+      "step": 1887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8210454788525277e-05,
+      "loss": 0.6243,
+      "step": 1888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8208322493391654e-05,
+      "loss": 0.66,
+      "step": 1889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8206189053661636e-05,
+      "loss": 0.7076,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8204054469632716e-05,
+      "loss": 0.6933,
+      "step": 1891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8201918741602548e-05,
+      "loss": 0.564,
+      "step": 1892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8199781869868948e-05,
+      "loss": 0.7176,
+      "step": 1893
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819764385472989e-05,
+      "loss": 0.7637,
+      "step": 1894
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.81955046964835e-05,
+      "loss": 0.8931,
+      "step": 1895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8193364395428078e-05,
+      "loss": 0.6746,
+      "step": 1896
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8191222951862066e-05,
+      "loss": 0.7115,
+      "step": 1897
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818908036608408e-05,
+      "loss": 1.127,
+      "step": 1898
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8186936638392885e-05,
+      "loss": 0.8347,
+      "step": 1899
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8184791769087415e-05,
+      "loss": 0.728,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818264575846675e-05,
+      "loss": 0.7611,
+      "step": 1901
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8180498606830146e-05,
+      "loss": 1.1175,
+      "step": 1902
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8178350314477006e-05,
+      "loss": 0.7082,
+      "step": 1903
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8176200881706892e-05,
+      "loss": 0.6983,
+      "step": 1904
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817405030881953e-05,
+      "loss": 1.1354,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8171898596114804e-05,
+      "loss": 0.5413,
+      "step": 1906
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8169745743892757e-05,
+      "loss": 0.6707,
+      "step": 1907
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8167591752453588e-05,
+      "loss": 0.8392,
+      "step": 1908
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816543662209766e-05,
+      "loss": 0.7948,
+      "step": 1909
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8163280353125486e-05,
+      "loss": 0.6276,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816112294583775e-05,
+      "loss": 0.9946,
+      "step": 1911
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8158964400535282e-05,
+      "loss": 0.6697,
+      "step": 1912
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8156804717519082e-05,
+      "loss": 0.7286,
+      "step": 1913
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8154643897090302e-05,
+      "loss": 1.1079,
+      "step": 1914
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8152481939550254e-05,
+      "loss": 0.5878,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815031884520041e-05,
+      "loss": 0.9531,
+      "step": 1916
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8148154614342393e-05,
+      "loss": 0.6934,
+      "step": 1917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8145989247277996e-05,
+      "loss": 0.7087,
+      "step": 1918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814382274430917e-05,
+      "loss": 0.9803,
+      "step": 1919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8141655105738006e-05,
+      "loss": 0.5845,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8139486331866783e-05,
+      "loss": 0.9821,
+      "step": 1921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813731642299791e-05,
+      "loss": 0.8941,
+      "step": 1922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813514537943397e-05,
+      "loss": 0.6863,
+      "step": 1923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8132973201477703e-05,
+      "loss": 0.777,
+      "step": 1924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8130799889432e-05,
+      "loss": 0.8418,
+      "step": 1925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8128625443599923e-05,
+      "loss": 1.2032,
+      "step": 1926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8126449864284678e-05,
+      "loss": 0.6203,
+      "step": 1927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8124273151789634e-05,
+      "loss": 0.6379,
+      "step": 1928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8122095306418326e-05,
+      "loss": 0.6723,
+      "step": 1929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8119916328474433e-05,
+      "loss": 0.7862,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8117736218261803e-05,
+      "loss": 0.7086,
+      "step": 1931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8115554976084438e-05,
+      "loss": 0.6108,
+      "step": 1932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8113372602246495e-05,
+      "loss": 0.8961,
+      "step": 1933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811118909705229e-05,
+      "loss": 0.7911,
+      "step": 1934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8109004460806306e-05,
+      "loss": 0.8805,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8106818693813174e-05,
+      "loss": 1.0641,
+      "step": 1936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8104631796377678e-05,
+      "loss": 0.8945,
+      "step": 1937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8102443768804775e-05,
+      "loss": 1.086,
+      "step": 1938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8100254611399563e-05,
+      "loss": 1.1655,
+      "step": 1939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8098064324467314e-05,
+      "loss": 0.7506,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.809587290831344e-05,
+      "loss": 0.5971,
+      "step": 1941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8093680363243528e-05,
+      "loss": 0.6938,
+      "step": 1942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8091486689563308e-05,
+      "loss": 0.7072,
+      "step": 1943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8089291887578678e-05,
+      "loss": 0.5918,
+      "step": 1944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8087095957595685e-05,
+      "loss": 0.9742,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8084898899920537e-05,
+      "loss": 0.811,
+      "step": 1946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80827007148596e-05,
+      "loss": 0.7616,
+      "step": 1947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8080501402719396e-05,
+      "loss": 0.8362,
+      "step": 1948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807830096380661e-05,
+      "loss": 0.7053,
+      "step": 1949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807609939842807e-05,
+      "loss": 0.6978,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8073896706890776e-05,
+      "loss": 0.707,
+      "step": 1951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807169288950187e-05,
+      "loss": 0.9486,
+      "step": 1952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8069487946568675e-05,
+      "loss": 0.7631,
+      "step": 1953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8067281878398642e-05,
+      "loss": 0.7171,
+      "step": 1954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80650746852994e-05,
+      "loss": 0.7162,
+      "step": 1955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8062866367578723e-05,
+      "loss": 0.769,
+      "step": 1956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8060656925544547e-05,
+      "loss": 0.7091,
+      "step": 1957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805844635950497e-05,
+      "loss": 0.6091,
+      "step": 1958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8056234669768235e-05,
+      "loss": 0.949,
+      "step": 1959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8054021856642747e-05,
+      "loss": 1.2068,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805180792043707e-05,
+      "loss": 0.6888,
+      "step": 1961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8049592861459926e-05,
+      "loss": 0.7226,
+      "step": 1962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8047376680020185e-05,
+      "loss": 0.677,
+      "step": 1963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8045159376426882e-05,
+      "loss": 0.6698,
+      "step": 1964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8042940950989205e-05,
+      "loss": 0.7192,
+      "step": 1965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8040721404016497e-05,
+      "loss": 0.7438,
+      "step": 1966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803850073581826e-05,
+      "loss": 0.7405,
+      "step": 1967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8036278946704153e-05,
+      "loss": 0.7838,
+      "step": 1968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.803405603698399e-05,
+      "loss": 0.6883,
+      "step": 1969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8031832006967737e-05,
+      "loss": 0.8194,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8029606856965527e-05,
+      "loss": 0.7847,
+      "step": 1971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8027380587287636e-05,
+      "loss": 0.7887,
+      "step": 1972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8025153198244507e-05,
+      "loss": 0.6703,
+      "step": 1973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8022924690146733e-05,
+      "loss": 0.7915,
+      "step": 1974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8020695063305062e-05,
+      "loss": 0.7591,
+      "step": 1975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8018464318030408e-05,
+      "loss": 0.689,
+      "step": 1976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8016232454633824e-05,
+      "loss": 0.6016,
+      "step": 1977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801399947342654e-05,
+      "loss": 0.814,
+      "step": 1978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.801176537471992e-05,
+      "loss": 0.952,
+      "step": 1979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8009530158825498e-05,
+      "loss": 0.8769,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800729382605496e-05,
+      "loss": 0.7939,
+      "step": 1981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800505637672015e-05,
+      "loss": 0.6091,
+      "step": 1982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8002817811133065e-05,
+      "loss": 0.764,
+      "step": 1983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8000578129605858e-05,
+      "loss": 0.745,
+      "step": 1984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7998337332450838e-05,
+      "loss": 0.6466,
+      "step": 1985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.799609541998047e-05,
+      "loss": 0.7015,
+      "step": 1986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7993852392507366e-05,
+      "loss": 0.7567,
+      "step": 1987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7991608250344312e-05,
+      "loss": 0.9249,
+      "step": 1988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7989362993804235e-05,
+      "loss": 1.0357,
+      "step": 1989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7987116623200218e-05,
+      "loss": 0.7293,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7984869138845505e-05,
+      "loss": 0.718,
+      "step": 1991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7982620541053497e-05,
+      "loss": 0.6802,
+      "step": 1992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798037083013774e-05,
+      "loss": 0.9187,
+      "step": 1993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7978120006411942e-05,
+      "loss": 0.7564,
+      "step": 1994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797586807018997e-05,
+      "loss": 0.7547,
+      "step": 1995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797361502178584e-05,
+      "loss": 0.6714,
+      "step": 1996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797136086151372e-05,
+      "loss": 0.7605,
+      "step": 1997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7969105589687944e-05,
+      "loss": 0.6639,
+      "step": 1998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796684920662299e-05,
+      "loss": 0.777,
+      "step": 1999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79645917126335e-05,
+      "loss": 0.6486,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7962333108034264e-05,
+      "loss": 0.6629,
+      "step": 2001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796007339314023e-05,
+      "loss": 0.7793,
+      "step": 2002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.79578125682665e-05,
+      "loss": 1.0077,
+      "step": 2003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7955550633728328e-05,
+      "loss": 0.6782,
+      "step": 2004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7953287589841135e-05,
+      "loss": 0.6387,
+      "step": 2005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795102343692048e-05,
+      "loss": 0.849,
+      "step": 2006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7948758175282083e-05,
+      "loss": 1.0145,
+      "step": 2007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7946491805241826e-05,
+      "loss": 0.6896,
+      "step": 2008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7944224327115734e-05,
+      "loss": 0.8775,
+      "step": 2009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7941955741219998e-05,
+      "loss": 0.7042,
+      "step": 2010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7939686047870947e-05,
+      "loss": 0.6339,
+      "step": 2011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7937415247385086e-05,
+      "loss": 0.8432,
+      "step": 2012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7935143340079054e-05,
+      "loss": 0.8684,
+      "step": 2013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793287032626966e-05,
+      "loss": 0.6034,
+      "step": 2014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793059620627386e-05,
+      "loss": 0.6775,
+      "step": 2015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.792832098040876e-05,
+      "loss": 1.085,
+      "step": 2016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792604464899163e-05,
+      "loss": 0.8249,
+      "step": 2017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792376721233989e-05,
+      "loss": 0.5694,
+      "step": 2018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792148867077111e-05,
+      "loss": 0.7228,
+      "step": 2019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7919209024603018e-05,
+      "loss": 0.7545,
+      "step": 2020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.79169282741535e-05,
+      "loss": 0.7916,
+      "step": 2021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7914646419740586e-05,
+      "loss": 0.6763,
+      "step": 2022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7912363461682467e-05,
+      "loss": 0.6944,
+      "step": 2023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791007940029749e-05,
+      "loss": 0.8329,
+      "step": 2024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790779423590415e-05,
+      "loss": 0.9626,
+      "step": 2025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7905507968821095e-05,
+      "loss": 0.6963,
+      "step": 2026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7903220599367134e-05,
+      "loss": 0.6734,
+      "step": 2027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7900932127861223e-05,
+      "loss": 0.7288,
+      "step": 2028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7898642554622476e-05,
+      "loss": 1.0649,
+      "step": 2029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789635187997016e-05,
+      "loss": 1.1567,
+      "step": 2030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789406010422369e-05,
+      "loss": 0.9761,
+      "step": 2031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.789176722770264e-05,
+      "loss": 0.9845,
+      "step": 2032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7889473250726745e-05,
+      "loss": 0.7312,
+      "step": 2033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7887178173615873e-05,
+      "loss": 0.8108,
+      "step": 2034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7884881996690066e-05,
+      "loss": 0.7834,
+      "step": 2035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78825847202695e-05,
+      "loss": 1.1269,
+      "step": 2036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.788028634467453e-05,
+      "loss": 0.671,
+      "step": 2037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787798687022564e-05,
+      "loss": 0.541,
+      "step": 2038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7875686297243476e-05,
+      "loss": 1.0087,
+      "step": 2039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.787338462604884e-05,
+      "loss": 0.6178,
+      "step": 2040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7871081856962688e-05,
+      "loss": 0.6422,
+      "step": 2041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7868777990306115e-05,
+      "loss": 1.0932,
+      "step": 2042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7866473026400396e-05,
+      "loss": 0.7588,
+      "step": 2043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7864166965566932e-05,
+      "loss": 1.0251,
+      "step": 2044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7861859808127288e-05,
+      "loss": 0.6349,
+      "step": 2045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7859551554403186e-05,
+      "loss": 0.7327,
+      "step": 2046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7857242204716497e-05,
+      "loss": 0.784,
+      "step": 2047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785493175938924e-05,
+      "loss": 0.7751,
+      "step": 2048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7852620218743596e-05,
+      "loss": 0.6722,
+      "step": 2049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7850307583101892e-05,
+      "loss": 0.8772,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7847993852786612e-05,
+      "loss": 0.7583,
+      "step": 2051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7845679028120386e-05,
+      "loss": 0.7543,
+      "step": 2052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7843363109426003e-05,
+      "loss": 0.7356,
+      "step": 2053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7841046097026404e-05,
+      "loss": 0.6844,
+      "step": 2054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7838727991244684e-05,
+      "loss": 1.0498,
+      "step": 2055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7836408792404075e-05,
+      "loss": 0.9309,
+      "step": 2056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783408850082799e-05,
+      "loss": 1.0526,
+      "step": 2057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783176711683997e-05,
+      "loss": 0.7642,
+      "step": 2058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7829444640763714e-05,
+      "loss": 0.7716,
+      "step": 2059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.782712107292308e-05,
+      "loss": 0.9339,
+      "step": 2060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7824796413642077e-05,
+      "loss": 0.6705,
+      "step": 2061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782247066324486e-05,
+      "loss": 0.6455,
+      "step": 2062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7820143822055735e-05,
+      "loss": 0.6739,
+      "step": 2063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7817815890399176e-05,
+      "loss": 0.9151,
+      "step": 2064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7815486868599786e-05,
+      "loss": 1.1733,
+      "step": 2065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7813156756982337e-05,
+      "loss": 0.6192,
+      "step": 2066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7810825555871752e-05,
+      "loss": 1.0473,
+      "step": 2067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7808493265593094e-05,
+      "loss": 0.5452,
+      "step": 2068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7806159886471588e-05,
+      "loss": 0.634,
+      "step": 2069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.780382541883261e-05,
+      "loss": 0.9366,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7801489863001686e-05,
+      "loss": 0.7999,
+      "step": 2071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779915321930449e-05,
+      "loss": 0.6813,
+      "step": 2072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7796815488066855e-05,
+      "loss": 1.0195,
+      "step": 2073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7794476669614766e-05,
+      "loss": 0.8719,
+      "step": 2074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779213676427435e-05,
+      "loss": 0.5721,
+      "step": 2075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778979577237189e-05,
+      "loss": 0.8088,
+      "step": 2076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7787453694233825e-05,
+      "loss": 0.6096,
+      "step": 2077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7785110530186742e-05,
+      "loss": 0.6463,
+      "step": 2078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778276628055738e-05,
+      "loss": 0.7101,
+      "step": 2079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778042094567263e-05,
+      "loss": 0.8463,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777807452585953e-05,
+      "loss": 0.9465,
+      "step": 2081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7775727021445275e-05,
+      "loss": 0.7688,
+      "step": 2082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.777337843275721e-05,
+      "loss": 0.807,
+      "step": 2083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7771028760122833e-05,
+      "loss": 0.8356,
+      "step": 2084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7768678003869783e-05,
+      "loss": 0.6968,
+      "step": 2085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7766326164325863e-05,
+      "loss": 0.6832,
+      "step": 2086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7763973241819017e-05,
+      "loss": 0.9629,
+      "step": 2087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776161923667735e-05,
+      "loss": 0.672,
+      "step": 2088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7759264149229105e-05,
+      "loss": 0.7364,
+      "step": 2089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775690797980269e-05,
+      "loss": 0.8016,
+      "step": 2090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7754550728726657e-05,
+      "loss": 0.719,
+      "step": 2091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.775219239632971e-05,
+      "loss": 0.6439,
+      "step": 2092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7749832982940693e-05,
+      "loss": 0.6361,
+      "step": 2093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7747472488888622e-05,
+      "loss": 0.7461,
+      "step": 2094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7745110914502646e-05,
+      "loss": 0.7853,
+      "step": 2095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7742748260112078e-05,
+      "loss": 1.0647,
+      "step": 2096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7740384526046366e-05,
+      "loss": 0.9622,
+      "step": 2097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773801971263512e-05,
+      "loss": 0.68,
+      "step": 2098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.77356538202081e-05,
+      "loss": 0.7363,
+      "step": 2099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7733286849095212e-05,
+      "loss": 0.797,
+      "step": 2100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7730918799626515e-05,
+      "loss": 0.7054,
+      "step": 2101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.772854967213222e-05,
+      "loss": 0.607,
+      "step": 2102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.772617946694268e-05,
+      "loss": 0.8571,
+      "step": 2103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772380818438841e-05,
+      "loss": 0.7577,
+      "step": 2104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.772143582480007e-05,
+      "loss": 1.0831,
+      "step": 2105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7719062388508467e-05,
+      "loss": 1.0738,
+      "step": 2106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771668787584456e-05,
+      "loss": 0.6405,
+      "step": 2107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7714312287139463e-05,
+      "loss": 0.8162,
+      "step": 2108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7711935622724438e-05,
+      "loss": 0.9071,
+      "step": 2109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7709557882930884e-05,
+      "loss": 0.7568,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7707179068090378e-05,
+      "loss": 0.6732,
+      "step": 2111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7704799178534614e-05,
+      "loss": 0.8034,
+      "step": 2112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.770241821459546e-05,
+      "loss": 0.7192,
+      "step": 2113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7700036176604925e-05,
+      "loss": 1.0596,
+      "step": 2114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7697653064895165e-05,
+      "loss": 0.7767,
+      "step": 2115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7695268879798498e-05,
+      "loss": 0.7465,
+      "step": 2116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769288362164737e-05,
+      "loss": 1.0414,
+      "step": 2117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.76904972907744e-05,
+      "loss": 0.6275,
+      "step": 2118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7688109887512346e-05,
+      "loss": 0.7203,
+      "step": 2119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768572141219411e-05,
+      "loss": 0.6579,
+      "step": 2120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7683331865152746e-05,
+      "loss": 0.9362,
+      "step": 2121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768094124672147e-05,
+      "loss": 1.1574,
+      "step": 2122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7678549557233636e-05,
+      "loss": 1.1325,
+      "step": 2123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7676156797022745e-05,
+      "loss": 1.0452,
+      "step": 2124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767376296642245e-05,
+      "loss": 0.7974,
+      "step": 2125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767136806576656e-05,
+      "loss": 0.685,
+      "step": 2126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766897209538903e-05,
+      "loss": 0.8594,
+      "step": 2127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7666575055623955e-05,
+      "loss": 0.6199,
+      "step": 2128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7664176946805595e-05,
+      "loss": 0.6209,
+      "step": 2129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766177776926834e-05,
+      "loss": 0.6817,
+      "step": 2130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7659377523346753e-05,
+      "loss": 0.716,
+      "step": 2131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7656976209375518e-05,
+      "loss": 1.0727,
+      "step": 2132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.765457382768949e-05,
+      "loss": 0.8052,
+      "step": 2133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7652170378623664e-05,
+      "loss": 0.7056,
+      "step": 2134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7649765862513192e-05,
+      "loss": 0.7579,
+      "step": 2135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7647360279693353e-05,
+      "loss": 1.0728,
+      "step": 2136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7644953630499602e-05,
+      "loss": 0.7161,
+      "step": 2137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7642545915267525e-05,
+      "loss": 0.7707,
+      "step": 2138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.764013713433286e-05,
+      "loss": 0.6705,
+      "step": 2139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7637727288031507e-05,
+      "loss": 0.7225,
+      "step": 2140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763531637669949e-05,
+      "loss": 0.7167,
+      "step": 2141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7632904400673e-05,
+      "loss": 0.7287,
+      "step": 2142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763049136028837e-05,
+      "loss": 0.7357,
+      "step": 2143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7628077255882086e-05,
+      "loss": 0.7002,
+      "step": 2144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7625662087790773e-05,
+      "loss": 0.7623,
+      "step": 2145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.762324585635122e-05,
+      "loss": 0.7883,
+      "step": 2146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.762082856190034e-05,
+      "loss": 0.7887,
+      "step": 2147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.761841020477522e-05,
+      "loss": 0.7209,
+      "step": 2148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7615990785313083e-05,
+      "loss": 0.7592,
+      "step": 2149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.76135703038513e-05,
+      "loss": 0.6618,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7611148760727384e-05,
+      "loss": 0.7064,
+      "step": 2151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7608726156279013e-05,
+      "loss": 0.9842,
+      "step": 2152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7606302490844e-05,
+      "loss": 0.8044,
+      "step": 2153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7603877764760308e-05,
+      "loss": 0.6438,
+      "step": 2154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760145197836605e-05,
+      "loss": 1.084,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7599025131999484e-05,
+      "loss": 0.6382,
+      "step": 2156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7596597225999018e-05,
+      "loss": 0.7346,
+      "step": 2157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7594168260703217e-05,
+      "loss": 0.6143,
+      "step": 2158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.759173823645077e-05,
+      "loss": 0.7896,
+      "step": 2159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758930715358053e-05,
+      "loss": 0.6296,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7586875012431504e-05,
+      "loss": 0.8379,
+      "step": 2161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758444181334283e-05,
+      "loss": 0.6519,
+      "step": 2162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758200755665381e-05,
+      "loss": 0.6596,
+      "step": 2163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7579572242703874e-05,
+      "loss": 0.7853,
+      "step": 2164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757713587183262e-05,
+      "loss": 0.6531,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757469844437978e-05,
+      "loss": 0.7362,
+      "step": 2166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757225996068524e-05,
+      "loss": 0.6989,
+      "step": 2167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756982042108902e-05,
+      "loss": 0.6792,
+      "step": 2168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7567379825931313e-05,
+      "loss": 0.6484,
+      "step": 2169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756493817555243e-05,
+      "loss": 0.7804,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7562495470292855e-05,
+      "loss": 0.6864,
+      "step": 2171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.75600517104932e-05,
+      "loss": 0.7557,
+      "step": 2172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.755760689649423e-05,
+      "loss": 0.7123,
+      "step": 2173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7555161028636863e-05,
+      "loss": 0.849,
+      "step": 2174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7552714107262158e-05,
+      "loss": 0.8522,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7550266132711317e-05,
+      "loss": 0.6843,
+      "step": 2176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7547817105325702e-05,
+      "loss": 0.7539,
+      "step": 2177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7545367025446807e-05,
+      "loss": 0.6418,
+      "step": 2178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7542915893416284e-05,
+      "loss": 0.7125,
+      "step": 2179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7540463709575927e-05,
+      "loss": 1.1682,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753801047426767e-05,
+      "loss": 0.6329,
+      "step": 2181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7535556187833606e-05,
+      "loss": 0.7429,
+      "step": 2182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7533100850615974e-05,
+      "loss": 0.8759,
+      "step": 2183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7530644462957145e-05,
+      "loss": 0.6123,
+      "step": 2184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752818702519965e-05,
+      "loss": 0.6055,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7525728537686165e-05,
+      "loss": 0.7102,
+      "step": 2186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7523269000759508e-05,
+      "loss": 0.816,
+      "step": 2187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.752080841476264e-05,
+      "loss": 0.6721,
+      "step": 2188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7518346780038683e-05,
+      "loss": 0.6201,
+      "step": 2189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7515884096930892e-05,
+      "loss": 1.1005,
+      "step": 2190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7513420365782666e-05,
+      "loss": 0.717,
+      "step": 2191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7510955586937565e-05,
+      "loss": 1.1576,
+      "step": 2192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.750848976073928e-05,
+      "loss": 0.7622,
+      "step": 2193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7506022887531657e-05,
+      "loss": 0.6895,
+      "step": 2194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7503554967658686e-05,
+      "loss": 0.7728,
+      "step": 2195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7501086001464495e-05,
+      "loss": 0.6132,
+      "step": 2196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7498615989293374e-05,
+      "loss": 0.804,
+      "step": 2197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7496144931489746e-05,
+      "loss": 0.7387,
+      "step": 2198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7493672828398187e-05,
+      "loss": 0.6971,
+      "step": 2199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749119968036341e-05,
+      "loss": 0.6625,
+      "step": 2200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748872548773028e-05,
+      "loss": 0.6868,
+      "step": 2201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7486250250843812e-05,
+      "loss": 0.7076,
+      "step": 2202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7483773970049158e-05,
+      "loss": 0.8968,
+      "step": 2203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.748129664569162e-05,
+      "loss": 0.7366,
+      "step": 2204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747881827811664e-05,
+      "loss": 0.8415,
+      "step": 2205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747633886766982e-05,
+      "loss": 0.7023,
+      "step": 2206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7473858414696883e-05,
+      "loss": 0.7588,
+      "step": 2207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7471376919543722e-05,
+      "loss": 0.6802,
+      "step": 2208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7468894382556368e-05,
+      "loss": 1.0202,
+      "step": 2209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7466410804080984e-05,
+      "loss": 0.8771,
+      "step": 2210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7463926184463896e-05,
+      "loss": 0.6032,
+      "step": 2211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7461440524051564e-05,
+      "loss": 0.609,
+      "step": 2212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.74589538231906e-05,
+      "loss": 0.5558,
+      "step": 2213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7456466082227758e-05,
+      "loss": 0.6441,
+      "step": 2214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7453977301509936e-05,
+      "loss": 0.7314,
+      "step": 2215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7451487481384177e-05,
+      "loss": 0.6682,
+      "step": 2216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7448996622197668e-05,
+      "loss": 1.1156,
+      "step": 2217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7446504724297752e-05,
+      "loss": 0.653,
+      "step": 2218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.74440117880319e-05,
+      "loss": 0.7719,
+      "step": 2219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7441517813747733e-05,
+      "loss": 0.7752,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7439022801793023e-05,
+      "loss": 0.6668,
+      "step": 2221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7436526752515684e-05,
+      "loss": 1.0444,
+      "step": 2222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743402966626378e-05,
+      "loss": 0.7103,
+      "step": 2223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7431531543385495e-05,
+      "loss": 0.6615,
+      "step": 2224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7429032384229194e-05,
+      "loss": 1.0555,
+      "step": 2225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742653218914336e-05,
+      "loss": 1.0246,
+      "step": 2226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7424030958476628e-05,
+      "loss": 0.7869,
+      "step": 2227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742152869257778e-05,
+      "loss": 1.0272,
+      "step": 2228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7419025391795742e-05,
+      "loss": 0.7358,
+      "step": 2229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7416521056479577e-05,
+      "loss": 0.6248,
+      "step": 2230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7414015686978503e-05,
+      "loss": 0.8432,
+      "step": 2231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.741150928364188e-05,
+      "loss": 0.8298,
+      "step": 2232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7409001846819205e-05,
+      "loss": 0.8914,
+      "step": 2233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.740649337686012e-05,
+      "loss": 0.5583,
+      "step": 2234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7403983874114422e-05,
+      "loss": 0.697,
+      "step": 2235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.740147333893204e-05,
+      "loss": 0.9234,
+      "step": 2236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739896177166305e-05,
+      "loss": 0.9375,
+      "step": 2237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7396449172657684e-05,
+      "loss": 0.679,
+      "step": 2238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7393935542266294e-05,
+      "loss": 0.7442,
+      "step": 2239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7391420880839397e-05,
+      "loss": 0.7762,
+      "step": 2240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7388905188727644e-05,
+      "loss": 0.9096,
+      "step": 2241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7386388466281833e-05,
+      "loss": 0.6998,
+      "step": 2242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7383870713852896e-05,
+      "loss": 0.8218,
+      "step": 2243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.738135193179193e-05,
+      "loss": 0.6048,
+      "step": 2244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7378832120450157e-05,
+      "loss": 0.6912,
+      "step": 2245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7376311280178944e-05,
+      "loss": 0.762,
+      "step": 2246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737378941132981e-05,
+      "loss": 0.635,
+      "step": 2247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7371266514254412e-05,
+      "loss": 0.6801,
+      "step": 2248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7368742589304552e-05,
+      "loss": 0.6545,
+      "step": 2249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7366217636832173e-05,
+      "loss": 0.7347,
+      "step": 2250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7363691657189367e-05,
+      "loss": 0.6712,
+      "step": 2251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736116465072836e-05,
+      "loss": 0.8535,
+      "step": 2252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7358636617801527e-05,
+      "loss": 0.8185,
+      "step": 2253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7356107558761385e-05,
+      "loss": 0.8329,
+      "step": 2254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7353577473960602e-05,
+      "loss": 0.6679,
+      "step": 2255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7351046363751973e-05,
+      "loss": 0.7469,
+      "step": 2256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734851422848845e-05,
+      "loss": 0.911,
+      "step": 2257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7345981068523116e-05,
+      "loss": 1.0883,
+      "step": 2258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734344688420921e-05,
+      "loss": 0.8497,
+      "step": 2259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7340911675900107e-05,
+      "loss": 0.7292,
+      "step": 2260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733837544394932e-05,
+      "loss": 0.6507,
+      "step": 2261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7335838188710516e-05,
+      "loss": 0.9345,
+      "step": 2262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7333299910537494e-05,
+      "loss": 0.7028,
+      "step": 2263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.73307606097842e-05,
+      "loss": 0.7277,
+      "step": 2264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7328220286804725e-05,
+      "loss": 0.8167,
+      "step": 2265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7325678941953298e-05,
+      "loss": 0.7282,
+      "step": 2266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7323136575584297e-05,
+      "loss": 0.7843,
+      "step": 2267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7320593188052237e-05,
+      "loss": 0.8717,
+      "step": 2268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7318048779711772e-05,
+      "loss": 0.7204,
+      "step": 2269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731550335091771e-05,
+      "loss": 0.8877,
+      "step": 2270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7312956902024987e-05,
+      "loss": 0.8375,
+      "step": 2271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731040943338869e-05,
+      "loss": 1.0161,
+      "step": 2272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730786094536405e-05,
+      "loss": 0.6804,
+      "step": 2273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7305311438306437e-05,
+      "loss": 0.6447,
+      "step": 2274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7302760912571363e-05,
+      "loss": 0.6877,
+      "step": 2275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7300209368514476e-05,
+      "loss": 0.7107,
+      "step": 2276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7297656806491577e-05,
+      "loss": 0.9944,
+      "step": 2277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7295103226858606e-05,
+      "loss": 0.6076,
+      "step": 2278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7292548629971636e-05,
+      "loss": 0.7058,
+      "step": 2279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7289993016186894e-05,
+      "loss": 0.743,
+      "step": 2280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728743638586074e-05,
+      "loss": 0.7729,
+      "step": 2281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728487873934969e-05,
+      "loss": 0.7862,
+      "step": 2282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7282320077010376e-05,
+      "loss": 0.7489,
+      "step": 2283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7279760399199588e-05,
+      "loss": 0.6889,
+      "step": 2284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727719970627427e-05,
+      "loss": 0.777,
+      "step": 2285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7274637998591482e-05,
+      "loss": 0.9641,
+      "step": 2286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.727207527650844e-05,
+      "loss": 0.6473,
+      "step": 2287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7269511540382495e-05,
+      "loss": 1.0298,
+      "step": 2288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726694679057115e-05,
+      "loss": 0.7862,
+      "step": 2289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7264381027432038e-05,
+      "loss": 0.8435,
+      "step": 2290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726181425132294e-05,
+      "loss": 1.1706,
+      "step": 2291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7259246462601774e-05,
+      "loss": 0.9054,
+      "step": 2292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.72566776616266e-05,
+      "loss": 0.6878,
+      "step": 2293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7254107848755626e-05,
+      "loss": 0.7903,
+      "step": 2294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725153702434719e-05,
+      "loss": 0.9381,
+      "step": 2295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7248965188759782e-05,
+      "loss": 0.6689,
+      "step": 2296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7246392342352018e-05,
+      "loss": 0.6617,
+      "step": 2297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7243818485482675e-05,
+      "loss": 1.0667,
+      "step": 2298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7241243618510656e-05,
+      "loss": 0.7737,
+      "step": 2299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7238667741795004e-05,
+      "loss": 0.6776,
+      "step": 2300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7236090855694915e-05,
+      "loss": 0.6944,
+      "step": 2301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7233512960569718e-05,
+      "loss": 0.6638,
+      "step": 2302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.723093405677888e-05,
+      "loss": 0.6287,
+      "step": 2303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7228354144682016e-05,
+      "loss": 0.662,
+      "step": 2304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7225773224638877e-05,
+      "loss": 1.0495,
+      "step": 2305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7223191297009352e-05,
+      "loss": 0.7181,
+      "step": 2306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7220608362153478e-05,
+      "loss": 0.7431,
+      "step": 2307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7218024420431423e-05,
+      "loss": 0.9651,
+      "step": 2308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7215439472203505e-05,
+      "loss": 0.72,
+      "step": 2309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7212853517830178e-05,
+      "loss": 0.728,
+      "step": 2310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7210266557672036e-05,
+      "loss": 0.633,
+      "step": 2311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720767859208981e-05,
+      "loss": 0.7313,
+      "step": 2312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720508962144438e-05,
+      "loss": 0.6949,
+      "step": 2313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7202499646096753e-05,
+      "loss": 1.1427,
+      "step": 2314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719990866640809e-05,
+      "loss": 0.5761,
+      "step": 2315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719731668273969e-05,
+      "loss": 0.695,
+      "step": 2316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7194723695452982e-05,
+      "loss": 1.0467,
+      "step": 2317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719212970490954e-05,
+      "loss": 0.6497,
+      "step": 2318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7189534711471082e-05,
+      "loss": 0.7058,
+      "step": 2319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.718693871549946e-05,
+      "loss": 1.0358,
+      "step": 2320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718434171735667e-05,
+      "loss": 0.8678,
+      "step": 2321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.718174371740485e-05,
+      "loss": 0.7191,
+      "step": 2322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7179144716006268e-05,
+      "loss": 0.7218,
+      "step": 2323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717654471352334e-05,
+      "loss": 0.6055,
+      "step": 2324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717394371031862e-05,
+      "loss": 0.6692,
+      "step": 2325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71713417067548e-05,
+      "loss": 0.7874,
+      "step": 2326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716873870319471e-05,
+      "loss": 0.6441,
+      "step": 2327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7166134700001324e-05,
+      "loss": 0.9717,
+      "step": 2328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7163529697537756e-05,
+      "loss": 0.6155,
+      "step": 2329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716092369616725e-05,
+      "loss": 0.7875,
+      "step": 2330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71583166962532e-05,
+      "loss": 0.9724,
+      "step": 2331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7155708698159132e-05,
+      "loss": 0.6661,
+      "step": 2332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7153099702248722e-05,
+      "loss": 0.6524,
+      "step": 2333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7150489708885764e-05,
+      "loss": 0.6309,
+      "step": 2334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7147878718434215e-05,
+      "loss": 0.6466,
+      "step": 2335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7145266731258153e-05,
+      "loss": 0.7958,
+      "step": 2336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7142653747721813e-05,
+      "loss": 0.7831,
+      "step": 2337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7140039768189547e-05,
+      "loss": 0.6974,
+      "step": 2338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7137424793025864e-05,
+      "loss": 0.6351,
+      "step": 2339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7134808822595402e-05,
+      "loss": 1.1338,
+      "step": 2340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7132191857262944e-05,
+      "loss": 0.7332,
+      "step": 2341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7129573897393402e-05,
+      "loss": 0.6349,
+      "step": 2342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712695494335184e-05,
+      "loss": 0.896,
+      "step": 2343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712433499550345e-05,
+      "loss": 1.0702,
+      "step": 2344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712171405421357e-05,
+      "loss": 0.9442,
+      "step": 2345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7119092119847666e-05,
+      "loss": 0.692,
+      "step": 2346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.711646919277136e-05,
+      "loss": 0.6952,
+      "step": 2347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7113845273350392e-05,
+      "loss": 0.7446,
+      "step": 2348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7111220361950654e-05,
+      "loss": 1.0204,
+      "step": 2349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7108594458938176e-05,
+      "loss": 0.9644,
+      "step": 2350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7105967564679117e-05,
+      "loss": 0.7817,
+      "step": 2351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7103339679539784e-05,
+      "loss": 0.745,
+      "step": 2352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7100710803886615e-05,
+      "loss": 0.7632,
+      "step": 2353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7098080938086194e-05,
+      "loss": 0.7481,
+      "step": 2354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7095450082505233e-05,
+      "loss": 0.709,
+      "step": 2355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7092818237510592e-05,
+      "loss": 0.6462,
+      "step": 2356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7090185403469264e-05,
+      "loss": 0.5947,
+      "step": 2357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7087551580748375e-05,
+      "loss": 0.7439,
+      "step": 2358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7084916769715204e-05,
+      "loss": 0.6333,
+      "step": 2359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7082280970737145e-05,
+      "loss": 0.7546,
+      "step": 2360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7079644184181753e-05,
+      "loss": 0.6763,
+      "step": 2361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.707700641041671e-05,
+      "loss": 0.6961,
+      "step": 2362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.707436764980983e-05,
+      "loss": 0.7508,
+      "step": 2363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7071727902729075e-05,
+      "loss": 1.0658,
+      "step": 2364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7069087169542538e-05,
+      "loss": 1.2408,
+      "step": 2365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7066445450618458e-05,
+      "loss": 0.8269,
+      "step": 2366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.70638027463252e-05,
+      "loss": 1.0436,
+      "step": 2367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706115905703127e-05,
+      "loss": 0.7886,
+      "step": 2368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7058514383105315e-05,
+      "loss": 0.6987,
+      "step": 2369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705586872491612e-05,
+      "loss": 0.7646,
+      "step": 2370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.70532220828326e-05,
+      "loss": 0.6051,
+      "step": 2371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7050574457223816e-05,
+      "loss": 0.7104,
+      "step": 2372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7047925848458956e-05,
+      "loss": 0.6005,
+      "step": 2373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7045276256907355e-05,
+      "loss": 0.7105,
+      "step": 2374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7042625682938484e-05,
+      "loss": 0.7091,
+      "step": 2375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7039974126921946e-05,
+      "loss": 0.8841,
+      "step": 2376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.703732158922748e-05,
+      "loss": 0.8129,
+      "step": 2377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7034668070224968e-05,
+      "loss": 0.66,
+      "step": 2378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7032013570284425e-05,
+      "loss": 0.7344,
+      "step": 2379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7029358089776004e-05,
+      "loss": 0.7699,
+      "step": 2380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.702670162906999e-05,
+      "loss": 0.7543,
+      "step": 2381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7024044188536814e-05,
+      "loss": 0.7328,
+      "step": 2382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7021385768547035e-05,
+      "loss": 0.8982,
+      "step": 2383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7018726369471358e-05,
+      "loss": 0.6292,
+      "step": 2384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701606599168061e-05,
+      "loss": 0.7163,
+      "step": 2385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701340463554577e-05,
+      "loss": 0.7246,
+      "step": 2386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7010742301437946e-05,
+      "loss": 1.0149,
+      "step": 2387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7008078989728377e-05,
+      "loss": 1.1708,
+      "step": 2388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7005414700788453e-05,
+      "loss": 0.7329,
+      "step": 2389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7002749434989684e-05,
+      "loss": 0.6469,
+      "step": 2390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7000083192703726e-05,
+      "loss": 0.6183,
+      "step": 2391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699741597430237e-05,
+      "loss": 0.7762,
+      "step": 2392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699474778015754e-05,
+      "loss": 0.7438,
+      "step": 2393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6992078610641304e-05,
+      "loss": 1.0592,
+      "step": 2394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6989408466125855e-05,
+      "loss": 0.6818,
+      "step": 2395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6986737346983526e-05,
+      "loss": 0.6548,
+      "step": 2396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6984065253586787e-05,
+      "loss": 1.1308,
+      "step": 2397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698139218630825e-05,
+      "loss": 0.6897,
+      "step": 2398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6978718145520646e-05,
+      "loss": 1.2078,
+      "step": 2399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697604313159686e-05,
+      "loss": 0.6776,
+      "step": 2400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6973367144909908e-05,
+      "loss": 0.6174,
+      "step": 2401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697069018583293e-05,
+      "loss": 0.9766,
+      "step": 2402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6968012254739213e-05,
+      "loss": 0.7691,
+      "step": 2403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.696533335200218e-05,
+      "loss": 0.8858,
+      "step": 2404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6962653477995384e-05,
+      "loss": 0.7096,
+      "step": 2405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.695997263309252e-05,
+      "loss": 0.6728,
+      "step": 2406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6957290817667406e-05,
+      "loss": 0.8561,
+      "step": 2407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.695460803209401e-05,
+      "loss": 1.1142,
+      "step": 2408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6951924276746425e-05,
+      "loss": 0.7791,
+      "step": 2409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6949239551998883e-05,
+      "loss": 0.6311,
+      "step": 2410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6946553858225755e-05,
+      "loss": 0.687,
+      "step": 2411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.694386719580154e-05,
+      "loss": 1.0461,
+      "step": 2412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6941179565100876e-05,
+      "loss": 0.6587,
+      "step": 2413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693849096649854e-05,
+      "loss": 0.8082,
+      "step": 2414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693580140036943e-05,
+      "loss": 0.6798,
+      "step": 2415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6933110867088593e-05,
+      "loss": 0.6777,
+      "step": 2416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693041936703121e-05,
+      "loss": 0.9063,
+      "step": 2417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6927726900572584e-05,
+      "loss": 0.7862,
+      "step": 2418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6925033468088172e-05,
+      "loss": 0.8297,
+      "step": 2419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6922339069953543e-05,
+      "loss": 0.7306,
+      "step": 2420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691964370654443e-05,
+      "loss": 0.795,
+      "step": 2421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6916947378236666e-05,
+      "loss": 0.7158,
+      "step": 2422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691425008540625e-05,
+      "loss": 0.7348,
+      "step": 2423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691155182842929e-05,
+      "loss": 0.6913,
+      "step": 2424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6908852607682052e-05,
+      "loss": 0.6774,
+      "step": 2425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6906152423540915e-05,
+      "loss": 0.6229,
+      "step": 2426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6903451276382406e-05,
+      "loss": 1.0698,
+      "step": 2427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6900749166583178e-05,
+      "loss": 0.6107,
+      "step": 2428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689804609452003e-05,
+      "loss": 0.5829,
+      "step": 2429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6895342060569885e-05,
+      "loss": 0.775,
+      "step": 2430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6892637065109796e-05,
+      "loss": 0.7256,
+      "step": 2431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6889931108516966e-05,
+      "loss": 1.1257,
+      "step": 2432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6887224191168715e-05,
+      "loss": 0.7592,
+      "step": 2433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688451631344251e-05,
+      "loss": 0.6036,
+      "step": 2434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6881807475715947e-05,
+      "loss": 0.6613,
+      "step": 2435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.687909767836675e-05,
+      "loss": 0.6556,
+      "step": 2436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6876386921772788e-05,
+      "loss": 0.7455,
+      "step": 2437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6873675206312058e-05,
+      "loss": 0.6127,
+      "step": 2438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6870962532362685e-05,
+      "loss": 0.7217,
+      "step": 2439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686824890030294e-05,
+      "loss": 0.7927,
+      "step": 2440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686553431051122e-05,
+      "loss": 0.9557,
+      "step": 2441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6862818763366048e-05,
+      "loss": 0.6281,
+      "step": 2442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6860102259246102e-05,
+      "loss": 0.7426,
+      "step": 2443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6857384798530176e-05,
+      "loss": 0.8928,
+      "step": 2444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68546663815972e-05,
+      "loss": 0.7092,
+      "step": 2445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6851947008826238e-05,
+      "loss": 1.1467,
+      "step": 2446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6849226680596488e-05,
+      "loss": 0.6693,
+      "step": 2447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6846505397287288e-05,
+      "loss": 0.7523,
+      "step": 2448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68437831592781e-05,
+      "loss": 1.1438,
+      "step": 2449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6841059966948518e-05,
+      "loss": 0.7844,
+      "step": 2450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6838335820678283e-05,
+      "loss": 1.1116,
+      "step": 2451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6835610720847247e-05,
+      "loss": 0.699,
+      "step": 2452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6832884667835415e-05,
+      "loss": 0.5727,
+      "step": 2453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6830157662022916e-05,
+      "loss": 0.6821,
+      "step": 2454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682742970379001e-05,
+      "loss": 0.5928,
+      "step": 2455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6824700793517096e-05,
+      "loss": 1.073,
+      "step": 2456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6821970931584705e-05,
+      "loss": 0.7022,
+      "step": 2457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681924011837349e-05,
+      "loss": 0.678,
+      "step": 2458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6816508354264252e-05,
+      "loss": 0.8045,
+      "step": 2459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6813775639637912e-05,
+      "loss": 0.7522,
+      "step": 2460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6811041974875536e-05,
+      "loss": 0.9956,
+      "step": 2461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680830736035831e-05,
+      "loss": 0.6525,
+      "step": 2462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680557179646756e-05,
+      "loss": 1.0051,
+      "step": 2463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680283528358474e-05,
+      "loss": 0.6725,
+      "step": 2464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6800097822091446e-05,
+      "loss": 0.8193,
+      "step": 2465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.679735941236939e-05,
+      "loss": 0.7737,
+      "step": 2466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6794620054800428e-05,
+      "loss": 0.8161,
+      "step": 2467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6791879749766545e-05,
+      "loss": 0.6733,
+      "step": 2468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6789138497649863e-05,
+      "loss": 0.9152,
+      "step": 2469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6786396298832622e-05,
+      "loss": 1.0372,
+      "step": 2470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6783653153697216e-05,
+      "loss": 0.7455,
+      "step": 2471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678090906262615e-05,
+      "loss": 0.7127,
+      "step": 2472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6778164026002068e-05,
+      "loss": 0.6883,
+      "step": 2473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6775418044207755e-05,
+      "loss": 1.001,
+      "step": 2474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677267111762611e-05,
+      "loss": 0.9265,
+      "step": 2475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.676992324664018e-05,
+      "loss": 0.7423,
+      "step": 2476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6767174431633143e-05,
+      "loss": 0.927,
+      "step": 2477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6764424672988294e-05,
+      "loss": 1.0615,
+      "step": 2478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6761673971089073e-05,
+      "loss": 1.0615,
+      "step": 2479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6758922326319043e-05,
+      "loss": 0.6554,
+      "step": 2480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6756169739061907e-05,
+      "loss": 0.5808,
+      "step": 2481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6753416209701497e-05,
+      "loss": 0.7072,
+      "step": 2482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6750661738621765e-05,
+      "loss": 0.6368,
+      "step": 2483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674790632620681e-05,
+      "loss": 0.8551,
+      "step": 2484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.674514997284086e-05,
+      "loss": 0.6974,
+      "step": 2485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6742392678908267e-05,
+      "loss": 0.8853,
+      "step": 2486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6739634444793516e-05,
+      "loss": 0.767,
+      "step": 2487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6736875270881222e-05,
+      "loss": 0.6424,
+      "step": 2488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6734115157556142e-05,
+      "loss": 0.9066,
+      "step": 2489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6731354105203147e-05,
+      "loss": 0.653,
+      "step": 2490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6728592114207254e-05,
+      "loss": 0.9423,
+      "step": 2491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67258291849536e-05,
+      "loss": 0.6492,
+      "step": 2492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6723065317827457e-05,
+      "loss": 0.8185,
+      "step": 2493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.672030051321423e-05,
+      "loss": 0.7591,
+      "step": 2494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6717534771499455e-05,
+      "loss": 0.6409,
+      "step": 2495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6714768093068792e-05,
+      "loss": 0.9071,
+      "step": 2496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.671200047830804e-05,
+      "loss": 0.8137,
+      "step": 2497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670923192760312e-05,
+      "loss": 1.0591,
+      "step": 2498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6706462441340093e-05,
+      "loss": 0.7053,
+      "step": 2499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.670369201990514e-05,
+      "loss": 0.9207,
+      "step": 2500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6700920663684584e-05,
+      "loss": 0.9944,
+      "step": 2501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.669814837306487e-05,
+      "loss": 1.0028,
+      "step": 2502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6695375148432574e-05,
+      "loss": 0.6452,
+      "step": 2503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6692600990174407e-05,
+      "loss": 0.7161,
+      "step": 2504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6689825898677202e-05,
+      "loss": 0.8038,
+      "step": 2505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6687049874327932e-05,
+      "loss": 0.6991,
+      "step": 2506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6684272917513694e-05,
+      "loss": 0.6047,
+      "step": 2507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6681495028621713e-05,
+      "loss": 0.6848,
+      "step": 2508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6678716208039353e-05,
+      "loss": 0.7501,
+      "step": 2509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6675936456154102e-05,
+      "loss": 0.7453,
+      "step": 2510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6673155773353575e-05,
+      "loss": 0.8876,
+      "step": 2511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6670374160025514e-05,
+      "loss": 0.7183,
+      "step": 2512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6667591616557808e-05,
+      "loss": 0.6827,
+      "step": 2513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6664808143338458e-05,
+      "loss": 0.7181,
+      "step": 2514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6662023740755602e-05,
+      "loss": 0.7094,
+      "step": 2515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6659238409197506e-05,
+      "loss": 0.6992,
+      "step": 2516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6656452149052568e-05,
+      "loss": 0.9524,
+      "step": 2517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6653664960709313e-05,
+      "loss": 0.7035,
+      "step": 2518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650876844556397e-05,
+      "loss": 0.7159,
+      "step": 2519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.66480878009826e-05,
+      "loss": 0.5931,
+      "step": 2520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6645297830376837e-05,
+      "loss": 0.7789,
+      "step": 2521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6642506933128156e-05,
+      "loss": 0.8763,
+      "step": 2522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6639715109625725e-05,
+      "loss": 0.658,
+      "step": 2523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6636922360258847e-05,
+      "loss": 0.6342,
+      "step": 2524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663412868541695e-05,
+      "loss": 0.7161,
+      "step": 2525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6631334085489596e-05,
+      "loss": 0.6498,
+      "step": 2526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6628538560866473e-05,
+      "loss": 0.6131,
+      "step": 2527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6625742111937397e-05,
+      "loss": 0.5966,
+      "step": 2528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6622944739092317e-05,
+      "loss": 0.7347,
+      "step": 2529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6620146442721305e-05,
+      "loss": 0.6728,
+      "step": 2530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6617347223214567e-05,
+      "loss": 0.7049,
+      "step": 2531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6614547080962435e-05,
+      "loss": 0.7378,
+      "step": 2532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661174601635537e-05,
+      "loss": 0.7196,
+      "step": 2533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6608944029783965e-05,
+      "loss": 0.6418,
+      "step": 2534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6606141121638938e-05,
+      "loss": 0.7501,
+      "step": 2535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660333729231113e-05,
+      "loss": 1.027,
+      "step": 2536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660053254219152e-05,
+      "loss": 0.6811,
+      "step": 2537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6597726871671214e-05,
+      "loss": 0.8523,
+      "step": 2538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6594920281141444e-05,
+      "loss": 0.7327,
+      "step": 2539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659211277099357e-05,
+      "loss": 0.6298,
+      "step": 2540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6589304341619075e-05,
+      "loss": 0.7509,
+      "step": 2541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6586494993409586e-05,
+      "loss": 0.7337,
+      "step": 2542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6583684726756845e-05,
+      "loss": 0.693,
+      "step": 2543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658087354205272e-05,
+      "loss": 0.673,
+      "step": 2544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578061439689212e-05,
+      "loss": 0.7701,
+      "step": 2545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6575248420058456e-05,
+      "loss": 0.7259,
+      "step": 2546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657243448355271e-05,
+      "loss": 0.9546,
+      "step": 2547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656961963056435e-05,
+      "loss": 0.6933,
+      "step": 2548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6566803861485898e-05,
+      "loss": 0.8063,
+      "step": 2549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656398717670999e-05,
+      "loss": 0.9602,
+      "step": 2550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6561169576629393e-05,
+      "loss": 0.6293,
+      "step": 2551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6558351061637005e-05,
+      "loss": 0.8303,
+      "step": 2552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6555531632125848e-05,
+      "loss": 0.6851,
+      "step": 2553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6552711288489076e-05,
+      "loss": 0.8194,
+      "step": 2554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654989003111996e-05,
+      "loss": 0.6052,
+      "step": 2555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6547067860411913e-05,
+      "loss": 0.6239,
+      "step": 2556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6544244776758466e-05,
+      "loss": 0.7495,
+      "step": 2557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654142078055328e-05,
+      "loss": 0.7698,
+      "step": 2558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6538595872190142e-05,
+      "loss": 0.6381,
+      "step": 2559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653577005206296e-05,
+      "loss": 0.7032,
+      "step": 2560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6532943320565792e-05,
+      "loss": 0.631,
+      "step": 2561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6530115678092792e-05,
+      "loss": 0.7137,
+      "step": 2562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6527287125038263e-05,
+      "loss": 0.6089,
+      "step": 2563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6524457661796626e-05,
+      "loss": 0.7535,
+      "step": 2564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.652162728876243e-05,
+      "loss": 0.6947,
+      "step": 2565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6518796006330356e-05,
+      "loss": 0.9153,
+      "step": 2566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65159638148952e-05,
+      "loss": 0.705,
+      "step": 2567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6513130714851902e-05,
+      "loss": 0.8228,
+      "step": 2568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6510296706595514e-05,
+      "loss": 0.8097,
+      "step": 2569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650746179052122e-05,
+      "loss": 0.768,
+      "step": 2570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650462596702433e-05,
+      "loss": 0.9197,
+      "step": 2571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6501789236500284e-05,
+      "loss": 0.7685,
+      "step": 2572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649895159934464e-05,
+      "loss": 0.6945,
+      "step": 2573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649611305595309e-05,
+      "loss": 0.7134,
+      "step": 2574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6493273606721453e-05,
+      "loss": 0.6611,
+      "step": 2575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649043325204567e-05,
+      "loss": 0.6245,
+      "step": 2576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.64875919923218e-05,
+      "loss": 0.8754,
+      "step": 2577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6484749827946058e-05,
+      "loss": 0.6991,
+      "step": 2578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648190675931475e-05,
+      "loss": 0.5869,
+      "step": 2579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6479062786824326e-05,
+      "loss": 1.0586,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647621791087136e-05,
+      "loss": 0.8291,
+      "step": 2581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6473372131852554e-05,
+      "loss": 0.7105,
+      "step": 2582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647052545016473e-05,
+      "loss": 0.7648,
+      "step": 2583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6467677866204835e-05,
+      "loss": 0.666,
+      "step": 2584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6464829380369955e-05,
+      "loss": 0.7091,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6461979993057286e-05,
+      "loss": 1.3074,
+      "step": 2586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645912970466416e-05,
+      "loss": 0.5921,
+      "step": 2587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6456278515588023e-05,
+      "loss": 0.7043,
+      "step": 2588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6453426426226466e-05,
+      "loss": 0.6382,
+      "step": 2589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.645057343697719e-05,
+      "loss": 0.6456,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644771954823802e-05,
+      "loss": 0.7798,
+      "step": 2591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6444864760406915e-05,
+      "loss": 0.6432,
+      "step": 2592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644200907388196e-05,
+      "loss": 0.7314,
+      "step": 2593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.643915248906136e-05,
+      "loss": 0.737,
+      "step": 2594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6436295006343443e-05,
+      "loss": 0.8118,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.643343662612667e-05,
+      "loss": 0.8727,
+      "step": 2596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6430577348809623e-05,
+      "loss": 0.5718,
+      "step": 2597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642771717479101e-05,
+      "loss": 0.9235,
+      "step": 2598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6424856104469658e-05,
+      "loss": 1.1324,
+      "step": 2599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642199413824453e-05,
+      "loss": 0.79,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6419131276514707e-05,
+      "loss": 0.9741,
+      "step": 2601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6416267519679396e-05,
+      "loss": 0.659,
+      "step": 2602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.641340286813793e-05,
+      "loss": 0.7381,
+      "step": 2603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6410537322289758e-05,
+      "loss": 0.6555,
+      "step": 2604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6407670882534473e-05,
+      "loss": 0.8611,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6404803549271777e-05,
+      "loss": 0.7147,
+      "step": 2606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6401935322901498e-05,
+      "loss": 0.6442,
+      "step": 2607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6399066203823592e-05,
+      "loss": 0.6492,
+      "step": 2608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6396196192438143e-05,
+      "loss": 0.868,
+      "step": 2609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6393325289145348e-05,
+      "loss": 0.6889,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.639045349434554e-05,
+      "loss": 1.0697,
+      "step": 2611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6387580808439176e-05,
+      "loss": 0.9761,
+      "step": 2612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.638470723182682e-05,
+      "loss": 0.8405,
+      "step": 2613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.638183276490919e-05,
+      "loss": 0.551,
+      "step": 2614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.63789574080871e-05,
+      "loss": 1.0149,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6376081161761502e-05,
+      "loss": 0.7845,
+      "step": 2616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637320402633347e-05,
+      "loss": 0.6493,
+      "step": 2617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6370326002204203e-05,
+      "loss": 0.7177,
+      "step": 2618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.636744708977502e-05,
+      "loss": 0.8967,
+      "step": 2619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6364567289447366e-05,
+      "loss": 0.7795,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6361686601622814e-05,
+      "loss": 0.7189,
+      "step": 2621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6358805026703053e-05,
+      "loss": 0.799,
+      "step": 2622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.63559225650899e-05,
+      "loss": 0.7231,
+      "step": 2623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.63530392171853e-05,
+      "loss": 0.6515,
+      "step": 2624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6350154983391312e-05,
+      "loss": 0.6604,
+      "step": 2625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634726986411012e-05,
+      "loss": 0.7083,
+      "step": 2626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6344383859744044e-05,
+      "loss": 0.6714,
+      "step": 2627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6341496970695514e-05,
+      "loss": 0.714,
+      "step": 2628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6338609197367087e-05,
+      "loss": 1.0023,
+      "step": 2629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.633572054016144e-05,
+      "loss": 0.9991,
+      "step": 2630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6332830999481387e-05,
+      "loss": 0.8448,
+      "step": 2631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6329940575729846e-05,
+      "loss": 0.7666,
+      "step": 2632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6327049269309872e-05,
+      "loss": 1.01,
+      "step": 2633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6324157080624642e-05,
+      "loss": 0.5832,
+      "step": 2634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6321264010077443e-05,
+      "loss": 0.9555,
+      "step": 2635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6318370058071705e-05,
+      "loss": 0.4882,
+      "step": 2636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6315475225010963e-05,
+      "loss": 0.7902,
+      "step": 2637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6312579511298885e-05,
+      "loss": 0.7371,
+      "step": 2638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.630968291733926e-05,
+      "loss": 0.6393,
+      "step": 2639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6306785443535996e-05,
+      "loss": 0.7274,
+      "step": 2640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.630388709029313e-05,
+      "loss": 0.6601,
+      "step": 2641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6300987858014817e-05,
+      "loss": 0.6988,
+      "step": 2642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6298087747105333e-05,
+      "loss": 0.6459,
+      "step": 2643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6295186757969087e-05,
+      "loss": 0.6991,
+      "step": 2644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6292284891010596e-05,
+      "loss": 0.6642,
+      "step": 2645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6289382146634506e-05,
+      "loss": 0.975,
+      "step": 2646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628647852524559e-05,
+      "loss": 0.7205,
+      "step": 2647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6283574027248735e-05,
+      "loss": 0.5796,
+      "step": 2648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6280668653048955e-05,
+      "loss": 0.6736,
+      "step": 2649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6277762403051386e-05,
+      "loss": 1.0943,
+      "step": 2650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6274855277661282e-05,
+      "loss": 1.1629,
+      "step": 2651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627194727728403e-05,
+      "loss": 0.9266,
+      "step": 2652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6269038402325126e-05,
+      "loss": 0.7177,
+      "step": 2653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6266128653190192e-05,
+      "loss": 0.8252,
+      "step": 2654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6263218030284978e-05,
+      "loss": 0.9918,
+      "step": 2655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6260306534015354e-05,
+      "loss": 0.8917,
+      "step": 2656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6257394164787297e-05,
+      "loss": 0.708,
+      "step": 2657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6254480923006924e-05,
+      "loss": 0.7597,
+      "step": 2658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625156680908047e-05,
+      "loss": 0.7212,
+      "step": 2659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624865182341429e-05,
+      "loss": 1.131,
+      "step": 2660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.624573596641485e-05,
+      "loss": 0.9879,
+      "step": 2661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6242819238488755e-05,
+      "loss": 0.5388,
+      "step": 2662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6239901640042726e-05,
+      "loss": 0.616,
+      "step": 2663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6236983171483593e-05,
+      "loss": 0.9592,
+      "step": 2664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6234063833218328e-05,
+      "loss": 0.7078,
+      "step": 2665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6231143625654007e-05,
+      "loss": 0.9012,
+      "step": 2666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.622822254919783e-05,
+      "loss": 1.0077,
+      "step": 2667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622530060425713e-05,
+      "loss": 0.7295,
+      "step": 2668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622237779123935e-05,
+      "loss": 0.6005,
+      "step": 2669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6219454110552057e-05,
+      "loss": 0.7979,
+      "step": 2670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.621652956260294e-05,
+      "loss": 0.5565,
+      "step": 2671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6213604147799803e-05,
+      "loss": 0.7762,
+      "step": 2672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6210677866550583e-05,
+      "loss": 0.9693,
+      "step": 2673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6207750719263327e-05,
+      "loss": 0.9039,
+      "step": 2674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.62048227063462e-05,
+      "loss": 0.638,
+      "step": 2675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6201893828207507e-05,
+      "loss": 0.8126,
+      "step": 2676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6198964085255652e-05,
+      "loss": 0.6513,
+      "step": 2677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.619603347789917e-05,
+      "loss": 0.7074,
+      "step": 2678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6193102006546712e-05,
+      "loss": 0.6357,
+      "step": 2679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6190169671607062e-05,
+      "loss": 0.8999,
+      "step": 2680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61872364734891e-05,
+      "loss": 0.6829,
+      "step": 2681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618430241260186e-05,
+      "loss": 0.7554,
+      "step": 2682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618136748935446e-05,
+      "loss": 0.6275,
+      "step": 2683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6178431704156167e-05,
+      "loss": 0.7435,
+      "step": 2684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.617549505741635e-05,
+      "loss": 1.0437,
+      "step": 2685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6172557549544504e-05,
+      "loss": 0.668,
+      "step": 2686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6169619180950253e-05,
+      "loss": 0.6969,
+      "step": 2687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6166679952043326e-05,
+      "loss": 0.9806,
+      "step": 2688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6163739863233586e-05,
+      "loss": 0.6829,
+      "step": 2689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6160798914931e-05,
+      "loss": 0.9142,
+      "step": 2690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.615785710754567e-05,
+      "loss": 1.0609,
+      "step": 2691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.615491444148781e-05,
+      "loss": 0.7197,
+      "step": 2692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6151970917167755e-05,
+      "loss": 0.7226,
+      "step": 2693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6149026534995962e-05,
+      "loss": 0.9762,
+      "step": 2694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6146081295383004e-05,
+      "loss": 0.9768,
+      "step": 2695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6143135198739575e-05,
+      "loss": 0.7438,
+      "step": 2696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6140188245476486e-05,
+      "loss": 0.7478,
+      "step": 2697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6137240436004678e-05,
+      "loss": 0.6947,
+      "step": 2698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.61342917707352e-05,
+      "loss": 0.6315,
+      "step": 2699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6131342250079218e-05,
+      "loss": 0.6911,
+      "step": 2700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.612839187444803e-05,
+      "loss": 1.0501,
+      "step": 2701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6125440644253045e-05,
+      "loss": 0.7972,
+      "step": 2702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.612248855990579e-05,
+      "loss": 0.7294,
+      "step": 2703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6119535621817918e-05,
+      "loss": 0.8686,
+      "step": 2704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6116581830401193e-05,
+      "loss": 1.0489,
+      "step": 2705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6113627186067505e-05,
+      "loss": 0.6806,
+      "step": 2706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6110671689228855e-05,
+      "loss": 0.6312,
+      "step": 2707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6107715340297373e-05,
+      "loss": 0.6329,
+      "step": 2708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6104758139685298e-05,
+      "loss": 0.6677,
+      "step": 2709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6101800087804993e-05,
+      "loss": 0.938,
+      "step": 2710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609884118506894e-05,
+      "loss": 0.8253,
+      "step": 2711
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609588143188974e-05,
+      "loss": 0.7381,
+      "step": 2712
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6092920828680106e-05,
+      "loss": 0.6699,
+      "step": 2713
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608995937585288e-05,
+      "loss": 0.9379,
+      "step": 2714
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6086997073821017e-05,
+      "loss": 0.7422,
+      "step": 2715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.608403392299758e-05,
+      "loss": 0.6713,
+      "step": 2716
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6081069923795775e-05,
+      "loss": 0.6089,
+      "step": 2717
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6078105076628907e-05,
+      "loss": 1.078,
+      "step": 2718
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.60751393819104e-05,
+      "loss": 0.8269,
+      "step": 2719
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6072172840053804e-05,
+      "loss": 0.8241,
+      "step": 2720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6069205451472788e-05,
+      "loss": 0.6583,
+      "step": 2721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6066237216581127e-05,
+      "loss": 0.728,
+      "step": 2722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6063268135792726e-05,
+      "loss": 0.703,
+      "step": 2723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6060298209521603e-05,
+      "loss": 0.5845,
+      "step": 2724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6057327438181897e-05,
+      "loss": 0.7662,
+      "step": 2725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6054355822187855e-05,
+      "loss": 0.648,
+      "step": 2726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6051383361953855e-05,
+      "loss": 0.8775,
+      "step": 2727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6048410057894388e-05,
+      "loss": 0.6693,
+      "step": 2728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6045435910424063e-05,
+      "loss": 0.661,
+      "step": 2729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6042460919957597e-05,
+      "loss": 0.7233,
+      "step": 2730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603948508690984e-05,
+      "loss": 0.7384,
+      "step": 2731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603650841169575e-05,
+      "loss": 0.6749,
+      "step": 2732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6033530894730403e-05,
+      "loss": 0.7266,
+      "step": 2733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6030552536428995e-05,
+      "loss": 0.6518,
+      "step": 2734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6027573337206847e-05,
+      "loss": 0.682,
+      "step": 2735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.602459329747937e-05,
+      "loss": 1.0926,
+      "step": 2736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6021612417662133e-05,
+      "loss": 1.0699,
+      "step": 2737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.601863069817078e-05,
+      "loss": 1.0999,
+      "step": 2738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6015648139421106e-05,
+      "loss": 0.7196,
+      "step": 2739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6012664741829004e-05,
+      "loss": 0.6003,
+      "step": 2740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6009680505810487e-05,
+      "loss": 0.6207,
+      "step": 2741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6006695431781686e-05,
+      "loss": 0.7705,
+      "step": 2742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.600370952015886e-05,
+      "loss": 1.0341,
+      "step": 2743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6000722771358364e-05,
+      "loss": 0.7096,
+      "step": 2744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5997735185796687e-05,
+      "loss": 0.5594,
+      "step": 2745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.599474676389042e-05,
+      "loss": 0.766,
+      "step": 2746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5991757506056288e-05,
+      "loss": 0.9862,
+      "step": 2747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.598876741271112e-05,
+      "loss": 0.5734,
+      "step": 2748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5985776484271863e-05,
+      "loss": 0.7907,
+      "step": 2749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5982784721155583e-05,
+      "loss": 0.7872,
+      "step": 2750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.597979212377946e-05,
+      "loss": 0.5998,
+      "step": 2751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5976798692560796e-05,
+      "loss": 0.675,
+      "step": 2752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5973804427917e-05,
+      "loss": 1.1349,
+      "step": 2753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.597080933026561e-05,
+      "loss": 0.6935,
+      "step": 2754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5967813400024264e-05,
+      "loss": 0.6042,
+      "step": 2755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5964816637610733e-05,
+      "loss": 0.7678,
+      "step": 2756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.596181904344289e-05,
+      "loss": 0.7798,
+      "step": 2757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.595882061793873e-05,
+      "loss": 0.6082,
+      "step": 2758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5955821361516367e-05,
+      "loss": 0.781,
+      "step": 2759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5952821274594022e-05,
+      "loss": 0.6761,
+      "step": 2760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.594982035759004e-05,
+      "loss": 0.6235,
+      "step": 2761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5946818610922886e-05,
+      "loss": 0.7638,
+      "step": 2762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5943816035011123e-05,
+      "loss": 0.7498,
+      "step": 2763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5940812630273446e-05,
+      "loss": 0.6993,
+      "step": 2764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593780839712866e-05,
+      "loss": 0.6884,
+      "step": 2765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5934803335995687e-05,
+      "loss": 0.6849,
+      "step": 2766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5931797447293553e-05,
+      "loss": 0.6814,
+      "step": 2767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5928790731441422e-05,
+      "loss": 0.9093,
+      "step": 2768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.592578318885855e-05,
+      "loss": 0.6846,
+      "step": 2769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.592277481996433e-05,
+      "loss": 0.6534,
+      "step": 2770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591976562517825e-05,
+      "loss": 0.5758,
+      "step": 2771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591675560491993e-05,
+      "loss": 1.0127,
+      "step": 2772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591374475960909e-05,
+      "loss": 0.6965,
+      "step": 2773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5910733089665577e-05,
+      "loss": 0.7271,
+      "step": 2774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5907720595509348e-05,
+      "loss": 1.1175,
+      "step": 2775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5904707277560475e-05,
+      "loss": 0.649,
+      "step": 2776
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5901693136239146e-05,
+      "loss": 0.6712,
+      "step": 2777
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5898678171965656e-05,
+      "loss": 1.0033,
+      "step": 2778
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5895662385160433e-05,
+      "loss": 0.7715,
+      "step": 2779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5892645776244002e-05,
+      "loss": 0.6413,
+      "step": 2780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5889628345637013e-05,
+      "loss": 0.8994,
+      "step": 2781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588661009376022e-05,
+      "loss": 0.6844,
+      "step": 2782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588359102103451e-05,
+      "loss": 0.7453,
+      "step": 2783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5880571127880863e-05,
+      "loss": 0.7548,
+      "step": 2784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587755041472038e-05,
+      "loss": 0.7811,
+      "step": 2785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587452888197429e-05,
+      "loss": 0.6994,
+      "step": 2786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5871506530063923e-05,
+      "loss": 0.5955,
+      "step": 2787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.586848335941072e-05,
+      "loss": 0.9374,
+      "step": 2788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.586545937043625e-05,
+      "loss": 0.7679,
+      "step": 2789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5862434563562183e-05,
+      "loss": 0.6143,
+      "step": 2790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585940893921031e-05,
+      "loss": 0.6806,
+      "step": 2791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5856382497802534e-05,
+      "loss": 1.0446,
+      "step": 2792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585335523976087e-05,
+      "loss": 0.7687,
+      "step": 2793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.585032716550746e-05,
+      "loss": 0.8213,
+      "step": 2794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5847298275464535e-05,
+      "loss": 1.0482,
+      "step": 2795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.584426857005446e-05,
+      "loss": 0.6896,
+      "step": 2796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5841238049699706e-05,
+      "loss": 0.8487,
+      "step": 2797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583820671482286e-05,
+      "loss": 0.7419,
+      "step": 2798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5835174565846624e-05,
+      "loss": 1.0885,
+      "step": 2799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5832141603193808e-05,
+      "loss": 0.6526,
+      "step": 2800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.582910782728734e-05,
+      "loss": 0.7072,
+      "step": 2801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5826073238550255e-05,
+      "loss": 1.0304,
+      "step": 2802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5823037837405715e-05,
+      "loss": 0.6803,
+      "step": 2803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.582000162427698e-05,
+      "loss": 0.6344,
+      "step": 2804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5816964599587436e-05,
+      "loss": 0.5912,
+      "step": 2805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581392676376057e-05,
+      "loss": 0.7437,
+      "step": 2806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.581088811721999e-05,
+      "loss": 0.7799,
+      "step": 2807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5807848660389416e-05,
+      "loss": 0.6905,
+      "step": 2808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580480839369268e-05,
+      "loss": 0.7146,
+      "step": 2809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5801767317553727e-05,
+      "loss": 0.6798,
+      "step": 2810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5798725432396615e-05,
+      "loss": 0.7344,
+      "step": 2811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5795682738645516e-05,
+      "loss": 0.6705,
+      "step": 2812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.579263923672471e-05,
+      "loss": 0.8017,
+      "step": 2813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.57895949270586e-05,
+      "loss": 0.6319,
+      "step": 2814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5786549810071688e-05,
+      "loss": 0.7541,
+      "step": 2815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5783503886188594e-05,
+      "loss": 0.8764,
+      "step": 2816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.578045715583406e-05,
+      "loss": 0.6337,
+      "step": 2817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577740961943293e-05,
+      "loss": 0.7432,
+      "step": 2818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577436127741016e-05,
+      "loss": 0.6034,
+      "step": 2819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5771312130190823e-05,
+      "loss": 1.0056,
+      "step": 2820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.57682621782001e-05,
+      "loss": 0.8289,
+      "step": 2821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.576521142186329e-05,
+      "loss": 0.6491,
+      "step": 2822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5762159861605798e-05,
+      "loss": 0.7104,
+      "step": 2823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5759107497853148e-05,
+      "loss": 1.0039,
+      "step": 2824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5756054331030966e-05,
+      "loss": 0.5728,
+      "step": 2825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5753000361565002e-05,
+      "loss": 0.6899,
+      "step": 2826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.574994558988111e-05,
+      "loss": 0.7525,
+      "step": 2827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5746890016405252e-05,
+      "loss": 0.869,
+      "step": 2828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5743833641563514e-05,
+      "loss": 0.7367,
+      "step": 2829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5740776465782087e-05,
+      "loss": 1.0766,
+      "step": 2830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5737718489487275e-05,
+      "loss": 0.6854,
+      "step": 2831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573465971310549e-05,
+      "loss": 0.6972,
+      "step": 2832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5731600137063257e-05,
+      "loss": 0.6354,
+      "step": 2833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5728539761787213e-05,
+      "loss": 1.0749,
+      "step": 2834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.572547858770411e-05,
+      "loss": 0.8091,
+      "step": 2835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.572241661524081e-05,
+      "loss": 0.7204,
+      "step": 2836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5719353844824284e-05,
+      "loss": 0.7016,
+      "step": 2837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5716290276881613e-05,
+      "loss": 0.738,
+      "step": 2838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5713225911839993e-05,
+      "loss": 0.9871,
+      "step": 2839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5710160750126725e-05,
+      "loss": 1.0842,
+      "step": 2840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5707094792169232e-05,
+      "loss": 0.6436,
+      "step": 2841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5704028038395035e-05,
+      "loss": 0.713,
+      "step": 2842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570096048923178e-05,
+      "loss": 0.6454,
+      "step": 2843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5697892145107212e-05,
+      "loss": 0.7888,
+      "step": 2844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569482300644919e-05,
+      "loss": 0.5875,
+      "step": 2845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5691753073685692e-05,
+      "loss": 0.6931,
+      "step": 2846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5688682347244794e-05,
+      "loss": 0.6321,
+      "step": 2847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5685610827554688e-05,
+      "loss": 0.7895,
+      "step": 2848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.568253851504368e-05,
+      "loss": 0.7989,
+      "step": 2849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5679465410140182e-05,
+      "loss": 0.8357,
+      "step": 2850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.567639151327272e-05,
+      "loss": 1.1167,
+      "step": 2851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5673316824869925e-05,
+      "loss": 0.6132,
+      "step": 2852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5670241345360547e-05,
+      "loss": 0.6939,
+      "step": 2853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5667165075173438e-05,
+      "loss": 0.7289,
+      "step": 2854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5664088014737565e-05,
+      "loss": 0.6267,
+      "step": 2855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5661010164482e-05,
+      "loss": 0.7612,
+      "step": 2856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5657931524835935e-05,
+      "loss": 0.7144,
+      "step": 2857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5654852096228662e-05,
+      "loss": 0.5969,
+      "step": 2858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565177187908959e-05,
+      "loss": 0.7154,
+      "step": 2859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564869087384823e-05,
+      "loss": 0.6339,
+      "step": 2860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564560908093421e-05,
+      "loss": 0.9676,
+      "step": 2861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5642526500777272e-05,
+      "loss": 0.9547,
+      "step": 2862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563944313380725e-05,
+      "loss": 0.5957,
+      "step": 2863
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5636358980454108e-05,
+      "loss": 0.7656,
+      "step": 2864
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.563327404114791e-05,
+      "loss": 1.0346,
+      "step": 2865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5630188316318822e-05,
+      "loss": 0.6422,
+      "step": 2866
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562710180639714e-05,
+      "loss": 0.8069,
+      "step": 2867
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5624014511813248e-05,
+      "loss": 1.0403,
+      "step": 2868
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5620926432997655e-05,
+      "loss": 0.799,
+      "step": 2869
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5617837570380968e-05,
+      "loss": 0.7274,
+      "step": 2870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5614747924393913e-05,
+      "loss": 0.7719,
+      "step": 2871
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.561165749546732e-05,
+      "loss": 0.6223,
+      "step": 2872
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5608566284032125e-05,
+      "loss": 0.7872,
+      "step": 2873
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5605474290519382e-05,
+      "loss": 0.709,
+      "step": 2874
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560238151536025e-05,
+      "loss": 0.8395,
+      "step": 2875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559928795898599e-05,
+      "loss": 0.5904,
+      "step": 2876
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5596193621827984e-05,
+      "loss": 0.7718,
+      "step": 2877
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5593098504317707e-05,
+      "loss": 0.7345,
+      "step": 2878
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5590002606886767e-05,
+      "loss": 0.9816,
+      "step": 2879
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5586905929966856e-05,
+      "loss": 0.6276,
+      "step": 2880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5583808473989788e-05,
+      "loss": 0.5704,
+      "step": 2881
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5580710239387485e-05,
+      "loss": 0.6333,
+      "step": 2882
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5577611226591974e-05,
+      "loss": 0.773,
+      "step": 2883
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5574511436035388e-05,
+      "loss": 0.8864,
+      "step": 2884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5571410868149975e-05,
+      "loss": 0.6657,
+      "step": 2885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5568309523368098e-05,
+      "loss": 0.6474,
+      "step": 2886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.55652074021222e-05,
+      "loss": 0.6183,
+      "step": 2887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556210450484487e-05,
+      "loss": 0.6841,
+      "step": 2888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5559000831968773e-05,
+      "loss": 0.6526,
+      "step": 2889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5555896383926703e-05,
+      "loss": 0.6379,
+      "step": 2890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5552791161151555e-05,
+      "loss": 0.5498,
+      "step": 2891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554968516407633e-05,
+      "loss": 0.9646,
+      "step": 2892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554657839313413e-05,
+      "loss": 0.6603,
+      "step": 2893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5543470848758193e-05,
+      "loss": 0.759,
+      "step": 2894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554036253138183e-05,
+      "loss": 0.8172,
+      "step": 2895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5537253441438484e-05,
+      "loss": 0.7166,
+      "step": 2896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5534143579361688e-05,
+      "loss": 0.699,
+      "step": 2897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5531032945585097e-05,
+      "loss": 0.8671,
+      "step": 2898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552792154054247e-05,
+      "loss": 1.0097,
+      "step": 2899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552480936466767e-05,
+      "loss": 0.8149,
+      "step": 2900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5521696418394672e-05,
+      "loss": 0.6473,
+      "step": 2901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5518582702157548e-05,
+      "loss": 0.7049,
+      "step": 2902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5515468216390494e-05,
+      "loss": 0.543,
+      "step": 2903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5512352961527803e-05,
+      "loss": 0.6923,
+      "step": 2904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.550923693800387e-05,
+      "loss": 0.7053,
+      "step": 2905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5506120146253213e-05,
+      "loss": 0.6818,
+      "step": 2906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.550300258671044e-05,
+      "loss": 0.8984,
+      "step": 2907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5499884259810284e-05,
+      "loss": 0.7184,
+      "step": 2908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5496765165987563e-05,
+      "loss": 0.6685,
+      "step": 2909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.549364530567722e-05,
+      "loss": 0.5775,
+      "step": 2910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5490524679314304e-05,
+      "loss": 0.7038,
+      "step": 2911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5487403287333954e-05,
+      "loss": 0.8519,
+      "step": 2912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5484281130171435e-05,
+      "loss": 0.7774,
+      "step": 2913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5481158208262117e-05,
+      "loss": 0.5731,
+      "step": 2914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547803452204146e-05,
+      "loss": 0.5982,
+      "step": 2915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5474910071945048e-05,
+      "loss": 0.835,
+      "step": 2916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5471784858408555e-05,
+      "loss": 0.8258,
+      "step": 2917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5468658881867785e-05,
+      "loss": 0.6187,
+      "step": 2918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546553214275863e-05,
+      "loss": 0.5592,
+      "step": 2919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546240464151709e-05,
+      "loss": 0.7824,
+      "step": 2920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5459276378579272e-05,
+      "loss": 0.8445,
+      "step": 2921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5456147354381402e-05,
+      "loss": 0.8141,
+      "step": 2922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5453017569359795e-05,
+      "loss": 0.6125,
+      "step": 2923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5449887023950877e-05,
+      "loss": 0.6488,
+      "step": 2924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5446755718591188e-05,
+      "loss": 0.8199,
+      "step": 2925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5443623653717357e-05,
+      "loss": 1.0514,
+      "step": 2926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5440490829766145e-05,
+      "loss": 0.9227,
+      "step": 2927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5437357247174393e-05,
+      "loss": 0.7107,
+      "step": 2928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5434222906379063e-05,
+      "loss": 0.9609,
+      "step": 2929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5431087807817212e-05,
+      "loss": 0.733,
+      "step": 2930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5427951951926016e-05,
+      "loss": 0.6558,
+      "step": 2931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5424815339142745e-05,
+      "loss": 0.785,
+      "step": 2932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5421677969904782e-05,
+      "loss": 1.0933,
+      "step": 2933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541853984464961e-05,
+      "loss": 0.7211,
+      "step": 2934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5415400963814818e-05,
+      "loss": 1.0494,
+      "step": 2935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.541226132783811e-05,
+      "loss": 0.7562,
+      "step": 2936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.540912093715728e-05,
+      "loss": 0.737,
+      "step": 2937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5405979792210238e-05,
+      "loss": 0.8568,
+      "step": 2938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5402837893434997e-05,
+      "loss": 0.9476,
+      "step": 2939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.539969524126967e-05,
+      "loss": 0.752,
+      "step": 2940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5396551836152477e-05,
+      "loss": 0.7546,
+      "step": 2941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5393407678521756e-05,
+      "loss": 0.7951,
+      "step": 2942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5390262768815933e-05,
+      "loss": 0.6043,
+      "step": 2943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5387117107473536e-05,
+      "loss": 0.5906,
+      "step": 2944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.538397069493322e-05,
+      "loss": 0.6311,
+      "step": 2945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5380823531633727e-05,
+      "loss": 0.8461,
+      "step": 2946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.537767561801391e-05,
+      "loss": 0.7748,
+      "step": 2947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.537452695451272e-05,
+      "loss": 0.7589,
+      "step": 2948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5371377541569218e-05,
+      "loss": 0.7836,
+      "step": 2949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5368227379622573e-05,
+      "loss": 0.8603,
+      "step": 2950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.536507646911205e-05,
+      "loss": 0.8772,
+      "step": 2951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5361924810477026e-05,
+      "loss": 0.607,
+      "step": 2952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5358772404156976e-05,
+      "loss": 0.7107,
+      "step": 2953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5355619250591483e-05,
+      "loss": 1.0411,
+      "step": 2954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5352465350220236e-05,
+      "loss": 0.7304,
+      "step": 2955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5349310703483027e-05,
+      "loss": 0.9806,
+      "step": 2956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5346155310819742e-05,
+      "loss": 0.6995,
+      "step": 2957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534299917267039e-05,
+      "loss": 0.757,
+      "step": 2958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5339842289475065e-05,
+      "loss": 1.0088,
+      "step": 2959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.533668466167398e-05,
+      "loss": 1.0685,
+      "step": 2960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5333526289707443e-05,
+      "loss": 0.7238,
+      "step": 2961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5330367174015866e-05,
+      "loss": 0.5573,
+      "step": 2962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.532720731503977e-05,
+      "loss": 0.6555,
+      "step": 2963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.532404671321977e-05,
+      "loss": 0.799,
+      "step": 2964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.53208853689966e-05,
+      "loss": 0.629,
+      "step": 2965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5317723282811088e-05,
+      "loss": 0.607,
+      "step": 2966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.531456045510416e-05,
+      "loss": 0.7291,
+      "step": 2967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5311396886316854e-05,
+      "loss": 0.773,
+      "step": 2968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.530823257689031e-05,
+      "loss": 0.7405,
+      "step": 2969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5305067527265768e-05,
+      "loss": 0.5348,
+      "step": 2970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5301901737884576e-05,
+      "loss": 0.9716,
+      "step": 2971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529873520918818e-05,
+      "loss": 0.7102,
+      "step": 2972
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529556794161813e-05,
+      "loss": 0.7273,
+      "step": 2973
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5292399935616087e-05,
+      "loss": 0.7345,
+      "step": 2974
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5289231191623804e-05,
+      "loss": 1.0623,
+      "step": 2975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5286061710083138e-05,
+      "loss": 0.6996,
+      "step": 2976
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528289149143606e-05,
+      "loss": 0.7258,
+      "step": 2977
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5279720536124632e-05,
+      "loss": 0.7566,
+      "step": 2978
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5276548844591022e-05,
+      "loss": 0.7586,
+      "step": 2979
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5273376417277503e-05,
+      "loss": 0.6947,
+      "step": 2980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5270203254626445e-05,
+      "loss": 1.0824,
+      "step": 2981
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5267029357080333e-05,
+      "loss": 0.8232,
+      "step": 2982
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.526385472508174e-05,
+      "loss": 0.6437,
+      "step": 2983
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5260679359073347e-05,
+      "loss": 0.6828,
+      "step": 2984
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5257503259497941e-05,
+      "loss": 0.6324,
+      "step": 2985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5254326426798406e-05,
+      "loss": 0.7723,
+      "step": 2986
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5251148861417733e-05,
+      "loss": 0.6324,
+      "step": 2987
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5247970563799009e-05,
+      "loss": 0.7875,
+      "step": 2988
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5244791534385428e-05,
+      "loss": 0.6069,
+      "step": 2989
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5241611773620283e-05,
+      "loss": 0.6233,
+      "step": 2990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5238431281946973e-05,
+      "loss": 0.8773,
+      "step": 2991
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5235250059808998e-05,
+      "loss": 0.6995,
+      "step": 2992
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5232068107649953e-05,
+      "loss": 0.7121,
+      "step": 2993
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.522888542591355e-05,
+      "loss": 0.7594,
+      "step": 2994
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5225702015043582e-05,
+      "loss": 0.6514,
+      "step": 2995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5222517875483963e-05,
+      "loss": 0.6675,
+      "step": 2996
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5219333007678691e-05,
+      "loss": 0.7801,
+      "step": 2997
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5216147412071883e-05,
+      "loss": 0.7045,
+      "step": 2998
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5212961089107747e-05,
+      "loss": 0.6504,
+      "step": 2999
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5209774039230594e-05,
+      "loss": 0.9778,
+      "step": 3000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5206586262884835e-05,
+      "loss": 0.8372,
+      "step": 3001
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.520339776051499e-05,
+      "loss": 0.7147,
+      "step": 3002
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.520020853256567e-05,
+      "loss": 0.7085,
+      "step": 3003
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5197018579481593e-05,
+      "loss": 0.6704,
+      "step": 3004
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5193827901707576e-05,
+      "loss": 0.7,
+      "step": 3005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5190636499688542e-05,
+      "loss": 0.6188,
+      "step": 3006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5187444373869506e-05,
+      "loss": 0.6266,
+      "step": 3007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5184251524695593e-05,
+      "loss": 0.67,
+      "step": 3008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.518105795261202e-05,
+      "loss": 0.83,
+      "step": 3009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5177863658064111e-05,
+      "loss": 0.643,
+      "step": 3010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5174668641497294e-05,
+      "loss": 0.6046,
+      "step": 3011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5171472903357085e-05,
+      "loss": 0.6146,
+      "step": 3012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5168276444089118e-05,
+      "loss": 1.017,
+      "step": 3013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5165079264139109e-05,
+      "loss": 0.8138,
+      "step": 3014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.516188136395289e-05,
+      "loss": 1.1145,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5158682743976385e-05,
+      "loss": 0.8125,
+      "step": 3016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5155483404655619e-05,
+      "loss": 0.757,
+      "step": 3017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.515228334643672e-05,
+      "loss": 0.8934,
+      "step": 3018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5149082569765915e-05,
+      "loss": 0.721,
+      "step": 3019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5145881075089533e-05,
+      "loss": 0.5554,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5142678862854e-05,
+      "loss": 0.6376,
+      "step": 3021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5139475933505839e-05,
+      "loss": 0.6315,
+      "step": 3022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5136272287491684e-05,
+      "loss": 0.6069,
+      "step": 3023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5133067925258258e-05,
+      "loss": 0.7395,
+      "step": 3024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5129862847252391e-05,
+      "loss": 0.6837,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.512665705392101e-05,
+      "loss": 0.7362,
+      "step": 3026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5123450545711136e-05,
+      "loss": 0.9359,
+      "step": 3027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5120243323069903e-05,
+      "loss": 1.0693,
+      "step": 3028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5117035386444534e-05,
+      "loss": 0.8338,
+      "step": 3029
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5113826736282353e-05,
+      "loss": 0.7759,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5110617373030786e-05,
+      "loss": 1.1742,
+      "step": 3031
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.510740729713736e-05,
+      "loss": 0.6245,
+      "step": 3032
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5104196509049696e-05,
+      "loss": 0.7209,
+      "step": 3033
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5100985009215519e-05,
+      "loss": 0.7256,
+      "step": 3034
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5097772798082649e-05,
+      "loss": 0.6461,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.509455987609901e-05,
+      "loss": 0.7529,
+      "step": 3036
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5091346243712619e-05,
+      "loss": 0.9609,
+      "step": 3037
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5088131901371604e-05,
+      "loss": 0.7049,
+      "step": 3038
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5084916849524173e-05,
+      "loss": 1.0246,
+      "step": 3039
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5081701088618656e-05,
+      "loss": 0.6535,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5078484619103457e-05,
+      "loss": 0.7581,
+      "step": 3041
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5075267441427101e-05,
+      "loss": 0.558,
+      "step": 3042
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5072049556038197e-05,
+      "loss": 0.7732,
+      "step": 3043
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5068830963385461e-05,
+      "loss": 0.6945,
+      "step": 3044
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5065611663917702e-05,
+      "loss": 0.747,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5062391658083831e-05,
+      "loss": 0.6612,
+      "step": 3046
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5059170946332855e-05,
+      "loss": 0.8393,
+      "step": 3047
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5055949529113884e-05,
+      "loss": 1.1008,
+      "step": 3048
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5052727406876125e-05,
+      "loss": 0.7126,
+      "step": 3049
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5049504580068877e-05,
+      "loss": 0.6051,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5046281049141543e-05,
+      "loss": 0.7635,
+      "step": 3051
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5043056814543624e-05,
+      "loss": 0.6506,
+      "step": 3052
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5039831876724716e-05,
+      "loss": 0.6751,
+      "step": 3053
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.503660623613452e-05,
+      "loss": 0.6399,
+      "step": 3054
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5033379893222826e-05,
+      "loss": 0.9023,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5030152848439526e-05,
+      "loss": 0.752,
+      "step": 3056
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5026925102234615e-05,
+      "loss": 0.6802,
+      "step": 3057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5023696655058175e-05,
+      "loss": 1.0705,
+      "step": 3058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5020467507360397e-05,
+      "loss": 1.0031,
+      "step": 3059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5017237659591561e-05,
+      "loss": 0.8524,
+      "step": 3060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.501400711220205e-05,
+      "loss": 0.7232,
+      "step": 3061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5010775865642338e-05,
+      "loss": 0.7376,
+      "step": 3062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5007543920363008e-05,
+      "loss": 0.7134,
+      "step": 3063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5004311276814724e-05,
+      "loss": 0.7928,
+      "step": 3064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5001077935448267e-05,
+      "loss": 1.0592,
+      "step": 3065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.49978438967145e-05,
+      "loss": 0.7516,
+      "step": 3066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.499460916106439e-05,
+      "loss": 0.7904,
+      "step": 3067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4991373728948997e-05,
+      "loss": 0.8631,
+      "step": 3068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4988137600819483e-05,
+      "loss": 0.7606,
+      "step": 3069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.49849007771271e-05,
+      "loss": 0.7611,
+      "step": 3070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4981663258323212e-05,
+      "loss": 0.9797,
+      "step": 3071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.497842504485926e-05,
+      "loss": 0.6118,
+      "step": 3072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4975186137186795e-05,
+      "loss": 0.6986,
+      "step": 3073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4971946535757457e-05,
+      "loss": 0.9911,
+      "step": 3074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4968706241022994e-05,
+      "loss": 0.9572,
+      "step": 3075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4965465253435243e-05,
+      "loss": 0.5957,
+      "step": 3076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4962223573446133e-05,
+      "loss": 0.849,
+      "step": 3077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4958981201507702e-05,
+      "loss": 0.8687,
+      "step": 3078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.495573813807207e-05,
+      "loss": 0.6604,
+      "step": 3079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4952494383591462e-05,
+      "loss": 0.7448,
+      "step": 3080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4949249938518203e-05,
+      "loss": 0.7771,
+      "step": 3081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4946004803304709e-05,
+      "loss": 0.9799,
+      "step": 3082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4942758978403484e-05,
+      "loss": 0.8453,
+      "step": 3083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4939512464267147e-05,
+      "loss": 0.7028,
+      "step": 3084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4936265261348398e-05,
+      "loss": 0.7291,
+      "step": 3085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4933017370100042e-05,
+      "loss": 0.9839,
+      "step": 3086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.492976879097497e-05,
+      "loss": 0.5803,
+      "step": 3087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4926519524426182e-05,
+      "loss": 0.7353,
+      "step": 3088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4923269570906758e-05,
+      "loss": 0.6756,
+      "step": 3089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.492001893086989e-05,
+      "loss": 0.6572,
+      "step": 3090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4916767604768855e-05,
+      "loss": 0.684,
+      "step": 3091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4913515593057031e-05,
+      "loss": 1.0015,
+      "step": 3092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4910262896187888e-05,
+      "loss": 0.8577,
+      "step": 3093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4907009514614992e-05,
+      "loss": 0.7698,
+      "step": 3094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.490375544879201e-05,
+      "loss": 1.0748,
+      "step": 3095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4900500699172693e-05,
+      "loss": 0.7066,
+      "step": 3096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4897245266210902e-05,
+      "loss": 1.0176,
+      "step": 3097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.489398915036058e-05,
+      "loss": 0.7162,
+      "step": 3098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4890732352075774e-05,
+      "loss": 0.6307,
+      "step": 3099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.488747487181062e-05,
+      "loss": 0.8122,
+      "step": 3100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4884216710019352e-05,
+      "loss": 0.7595,
+      "step": 3101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4880957867156305e-05,
+      "loss": 0.7946,
+      "step": 3102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.48776983436759e-05,
+      "loss": 0.6723,
+      "step": 3103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4874438140032652e-05,
+      "loss": 0.7212,
+      "step": 3104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4871177256681181e-05,
+      "loss": 1.0076,
+      "step": 3105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4867915694076191e-05,
+      "loss": 0.5888,
+      "step": 3106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4864653452672488e-05,
+      "loss": 1.018,
+      "step": 3107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4861390532924969e-05,
+      "loss": 0.8087,
+      "step": 3108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4858126935288627e-05,
+      "loss": 0.9939,
+      "step": 3109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4854862660218547e-05,
+      "loss": 0.6024,
+      "step": 3110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4851597708169916e-05,
+      "loss": 0.5935,
+      "step": 3111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4848332079598004e-05,
+      "loss": 0.9298,
+      "step": 3112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4845065774958188e-05,
+      "loss": 0.6575,
+      "step": 3113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4841798794705924e-05,
+      "loss": 1.033,
+      "step": 3114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4838531139296779e-05,
+      "loss": 0.899,
+      "step": 3115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.48352628091864e-05,
+      "loss": 0.6986,
+      "step": 3116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4831993804830537e-05,
+      "loss": 0.6636,
+      "step": 3117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4828724126685032e-05,
+      "loss": 0.993,
+      "step": 3118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4825453775205817e-05,
+      "loss": 0.6925,
+      "step": 3119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4822182750848923e-05,
+      "loss": 0.659,
+      "step": 3120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4818911054070473e-05,
+      "loss": 0.6181,
+      "step": 3121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4815638685326683e-05,
+      "loss": 0.6275,
+      "step": 3122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.481236564507386e-05,
+      "loss": 0.6402,
+      "step": 3123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4809091933768415e-05,
+      "loss": 1.0046,
+      "step": 3124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4805817551866839e-05,
+      "loss": 0.9659,
+      "step": 3125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4802542499825727e-05,
+      "loss": 0.9379,
+      "step": 3126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.479926677810176e-05,
+      "loss": 0.7969,
+      "step": 3127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4795990387151719e-05,
+      "loss": 0.7191,
+      "step": 3128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.479271332743247e-05,
+      "loss": 0.9139,
+      "step": 3129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4789435599400985e-05,
+      "loss": 0.6509,
+      "step": 3130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4786157203514316e-05,
+      "loss": 0.9533,
+      "step": 3131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4782878140229615e-05,
+      "loss": 0.6533,
+      "step": 3132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4779598410004124e-05,
+      "loss": 0.6523,
+      "step": 3133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4776318013295185e-05,
+      "loss": 0.9609,
+      "step": 3134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4773036950560222e-05,
+      "loss": 0.9041,
+      "step": 3135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.476975522225676e-05,
+      "loss": 0.8625,
+      "step": 3136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4766472828842415e-05,
+      "loss": 0.6458,
+      "step": 3137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4763189770774892e-05,
+      "loss": 0.8581,
+      "step": 3138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4759906048511998e-05,
+      "loss": 0.6416,
+      "step": 3139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4756621662511618e-05,
+      "loss": 0.7564,
+      "step": 3140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4753336613231745e-05,
+      "loss": 0.95,
+      "step": 3141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4750050901130454e-05,
+      "loss": 0.76,
+      "step": 3142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.474676452666592e-05,
+      "loss": 0.8741,
+      "step": 3143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4743477490296398e-05,
+      "loss": 0.7669,
+      "step": 3144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4740189792480251e-05,
+      "loss": 1.0838,
+      "step": 3145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4736901433675923e-05,
+      "loss": 0.966,
+      "step": 3146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4733612414341955e-05,
+      "loss": 0.911,
+      "step": 3147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.473032273493698e-05,
+      "loss": 0.6968,
+      "step": 3148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.472703239591972e-05,
+      "loss": 0.9058,
+      "step": 3149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4723741397748993e-05,
+      "loss": 0.6565,
+      "step": 3150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4720449740883706e-05,
+      "loss": 0.8218,
+      "step": 3151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.471715742578286e-05,
+      "loss": 0.7443,
+      "step": 3152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4713864452905547e-05,
+      "loss": 0.5846,
+      "step": 3153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4710570822710946e-05,
+      "loss": 1.151,
+      "step": 3154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4707276535658337e-05,
+      "loss": 0.6031,
+      "step": 3155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4703981592207081e-05,
+      "loss": 0.7821,
+      "step": 3156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4700685992816644e-05,
+      "loss": 0.5824,
+      "step": 3157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4697389737946568e-05,
+      "loss": 0.6551,
+      "step": 3158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.46940928280565e-05,
+      "loss": 0.7422,
+      "step": 3159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4690795263606168e-05,
+      "loss": 0.6525,
+      "step": 3160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4687497045055398e-05,
+      "loss": 0.5972,
+      "step": 3161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4684198172864105e-05,
+      "loss": 0.8689,
+      "step": 3162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4680898647492295e-05,
+      "loss": 0.6313,
+      "step": 3163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4677598469400058e-05,
+      "loss": 0.7625,
+      "step": 3164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4674297639047595e-05,
+      "loss": 0.8129,
+      "step": 3165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4670996156895174e-05,
+      "loss": 0.8167,
+      "step": 3166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.466769402340317e-05,
+      "loss": 0.6341,
+      "step": 3167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4664391239032044e-05,
+      "loss": 0.7103,
+      "step": 3168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4661087804242342e-05,
+      "loss": 0.7557,
+      "step": 3169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4657783719494716e-05,
+      "loss": 0.5792,
+      "step": 3170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.465447898524989e-05,
+      "loss": 1.0007,
+      "step": 3171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4651173601968691e-05,
+      "loss": 0.662,
+      "step": 3172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4647867570112031e-05,
+      "loss": 0.8034,
+      "step": 3173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4644560890140918e-05,
+      "loss": 0.7673,
+      "step": 3174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.464125356251644e-05,
+      "loss": 0.6375,
+      "step": 3175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4637945587699791e-05,
+      "loss": 1.0716,
+      "step": 3176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4634636966152238e-05,
+      "loss": 0.7977,
+      "step": 3177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4631327698335155e-05,
+      "loss": 0.8131,
+      "step": 3178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4628017784709989e-05,
+      "loss": 0.6784,
+      "step": 3179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4624707225738292e-05,
+      "loss": 0.7481,
+      "step": 3180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4621396021881695e-05,
+      "loss": 0.7624,
+      "step": 3181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4618084173601928e-05,
+      "loss": 0.7713,
+      "step": 3182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4614771681360799e-05,
+      "loss": 0.7482,
+      "step": 3183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.461145854562022e-05,
+      "loss": 0.609,
+      "step": 3184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4608144766842185e-05,
+      "loss": 0.8572,
+      "step": 3185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4604830345488776e-05,
+      "loss": 0.5824,
+      "step": 3186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4601515282022171e-05,
+      "loss": 0.8357,
+      "step": 3187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.459819957690463e-05,
+      "loss": 0.8177,
+      "step": 3188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.459488323059851e-05,
+      "loss": 0.6273,
+      "step": 3189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4591566243566245e-05,
+      "loss": 0.9604,
+      "step": 3190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4588248616270378e-05,
+      "loss": 0.9143,
+      "step": 3191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4584930349173522e-05,
+      "loss": 0.7418,
+      "step": 3192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4581611442738392e-05,
+      "loss": 0.774,
+      "step": 3193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4578291897427787e-05,
+      "loss": 0.9081,
+      "step": 3194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4574971713704597e-05,
+      "loss": 0.7706,
+      "step": 3195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4571650892031794e-05,
+      "loss": 1.1901,
+      "step": 3196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4568329432872452e-05,
+      "loss": 0.6134,
+      "step": 3197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4565007336689721e-05,
+      "loss": 0.558,
+      "step": 3198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4561684603946849e-05,
+      "loss": 0.6447,
+      "step": 3199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4558361235107165e-05,
+      "loss": 0.7782,
+      "step": 3200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4555037230634097e-05,
+      "loss": 0.6297,
+      "step": 3201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4551712590991146e-05,
+      "loss": 0.7769,
+      "step": 3202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4548387316641921e-05,
+      "loss": 0.617,
+      "step": 3203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4545061408050107e-05,
+      "loss": 0.7008,
+      "step": 3204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4541734865679476e-05,
+      "loss": 0.7112,
+      "step": 3205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4538407689993896e-05,
+      "loss": 0.5789,
+      "step": 3206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4535079881457317e-05,
+      "loss": 0.625,
+      "step": 3207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4531751440533784e-05,
+      "loss": 0.7826,
+      "step": 3208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.452842236768742e-05,
+      "loss": 0.7411,
+      "step": 3209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.452509266338245e-05,
+      "loss": 0.7393,
+      "step": 3210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.452176232808317e-05,
+      "loss": 0.7921,
+      "step": 3211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4518431362253982e-05,
+      "loss": 0.9413,
+      "step": 3212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4515099766359363e-05,
+      "loss": 0.7485,
+      "step": 3213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4511767540863881e-05,
+      "loss": 0.6964,
+      "step": 3214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4508434686232195e-05,
+      "loss": 0.9985,
+      "step": 3215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.450510120292905e-05,
+      "loss": 0.886,
+      "step": 3216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4501767091419279e-05,
+      "loss": 0.8016,
+      "step": 3217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.44984323521678e-05,
+      "loss": 0.8325,
+      "step": 3218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4495096985639618e-05,
+      "loss": 0.7237,
+      "step": 3219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4491760992299831e-05,
+      "loss": 0.8201,
+      "step": 3220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.448842437261362e-05,
+      "loss": 0.7141,
+      "step": 3221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4485087127046256e-05,
+      "loss": 0.8117,
+      "step": 3222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4481749256063096e-05,
+      "loss": 0.9036,
+      "step": 3223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4478410760129582e-05,
+      "loss": 0.6884,
+      "step": 3224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4475071639711247e-05,
+      "loss": 0.773,
+      "step": 3225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.447173189527371e-05,
+      "loss": 0.817,
+      "step": 3226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4468391527282672e-05,
+      "loss": 0.9483,
+      "step": 3227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4465050536203932e-05,
+      "loss": 1.1052,
+      "step": 3228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4461708922503364e-05,
+      "loss": 0.5895,
+      "step": 3229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4458366686646934e-05,
+      "loss": 0.6502,
+      "step": 3230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4455023829100698e-05,
+      "loss": 0.8681,
+      "step": 3231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.445168035033079e-05,
+      "loss": 0.6913,
+      "step": 3232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4448336250803446e-05,
+      "loss": 0.6525,
+      "step": 3233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4444991530984967e-05,
+      "loss": 0.8346,
+      "step": 3234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4441646191341758e-05,
+      "loss": 0.7771,
+      "step": 3235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4438300232340306e-05,
+      "loss": 0.5593,
+      "step": 3236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4434953654447179e-05,
+      "loss": 0.9478,
+      "step": 3237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4431606458129037e-05,
+      "loss": 0.7745,
+      "step": 3238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4428258643852624e-05,
+      "loss": 0.9082,
+      "step": 3239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.442491021208477e-05,
+      "loss": 0.5479,
+      "step": 3240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4421561163292397e-05,
+      "loss": 0.9414,
+      "step": 3241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4418211497942501e-05,
+      "loss": 0.5904,
+      "step": 3242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4414861216502173e-05,
+      "loss": 0.6207,
+      "step": 3243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.441151031943859e-05,
+      "loss": 0.8212,
+      "step": 3244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4408158807219008e-05,
+      "loss": 0.8044,
+      "step": 3245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4404806680310774e-05,
+      "loss": 0.9837,
+      "step": 3246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4401453939181325e-05,
+      "loss": 0.6691,
+      "step": 3247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4398100584298174e-05,
+      "loss": 0.7682,
+      "step": 3248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4394746616128924e-05,
+      "loss": 1.0251,
+      "step": 3249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4391392035141265e-05,
+      "loss": 0.6455,
+      "step": 3250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4388036841802973e-05,
+      "loss": 0.5987,
+      "step": 3251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4384681036581904e-05,
+      "loss": 0.86,
+      "step": 3252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4381324619946007e-05,
+      "loss": 0.657,
+      "step": 3253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4377967592363308e-05,
+      "loss": 0.6881,
+      "step": 3254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4374609954301926e-05,
+      "loss": 0.5278,
+      "step": 3255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.437125170623006e-05,
+      "loss": 0.6637,
+      "step": 3256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4367892848615992e-05,
+      "loss": 0.7331,
+      "step": 3257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4364533381928096e-05,
+      "loss": 0.7913,
+      "step": 3258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4361173306634828e-05,
+      "loss": 0.6496,
+      "step": 3259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.435781262320473e-05,
+      "loss": 0.992,
+      "step": 3260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4354451332106423e-05,
+      "loss": 0.7927,
+      "step": 3261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4351089433808619e-05,
+      "loss": 0.6663,
+      "step": 3262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434772692878011e-05,
+      "loss": 0.754,
+      "step": 3263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434436381748978e-05,
+      "loss": 0.7253,
+      "step": 3264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4341000100406586e-05,
+      "loss": 0.7276,
+      "step": 3265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4337635777999583e-05,
+      "loss": 0.7775,
+      "step": 3266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4334270850737899e-05,
+      "loss": 1.1601,
+      "step": 3267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4330905319090753e-05,
+      "loss": 0.7172,
+      "step": 3268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4327539183527447e-05,
+      "loss": 1.023,
+      "step": 3269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4324172444517364e-05,
+      "loss": 0.6789,
+      "step": 3270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4320805102529973e-05,
+      "loss": 0.6386,
+      "step": 3271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.431743715803483e-05,
+      "loss": 0.6711,
+      "step": 3272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4314068611501572e-05,
+      "loss": 0.818,
+      "step": 3273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.431069946339992e-05,
+      "loss": 0.849,
+      "step": 3274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.430732971419968e-05,
+      "loss": 0.704,
+      "step": 3275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4303959364370741e-05,
+      "loss": 1.0467,
+      "step": 3276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4300588414383077e-05,
+      "loss": 0.636,
+      "step": 3277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4297216864706742e-05,
+      "loss": 0.7149,
+      "step": 3278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.429384471581188e-05,
+      "loss": 0.8866,
+      "step": 3279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4290471968168712e-05,
+      "loss": 0.8615,
+      "step": 3280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4287098622247549e-05,
+      "loss": 0.6139,
+      "step": 3281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4283724678518776e-05,
+      "loss": 0.6598,
+      "step": 3282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4280350137452872e-05,
+      "loss": 0.7067,
+      "step": 3283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4276974999520392e-05,
+      "loss": 0.7244,
+      "step": 3284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4273599265191978e-05,
+      "loss": 0.6845,
+      "step": 3285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4270222934938351e-05,
+      "loss": 0.8107,
+      "step": 3286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4266846009230326e-05,
+      "loss": 0.6364,
+      "step": 3287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4263468488538782e-05,
+      "loss": 0.7367,
+      "step": 3288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4260090373334703e-05,
+      "loss": 0.6075,
+      "step": 3289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4256711664089135e-05,
+      "loss": 0.6895,
+      "step": 3290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4253332361273224e-05,
+      "loss": 0.6441,
+      "step": 3291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4249952465358188e-05,
+      "loss": 0.8141,
+      "step": 3292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4246571976815333e-05,
+      "loss": 0.674,
+      "step": 3293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4243190896116042e-05,
+      "loss": 0.5757,
+      "step": 3294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4239809223731786e-05,
+      "loss": 1.1004,
+      "step": 3295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.423642696013412e-05,
+      "loss": 0.7667,
+      "step": 3296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4233044105794675e-05,
+      "loss": 0.6883,
+      "step": 3297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4229660661185171e-05,
+      "loss": 0.7025,
+      "step": 3298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4226276626777402e-05,
+      "loss": 0.622,
+      "step": 3299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4222892003043254e-05,
+      "loss": 0.6996,
+      "step": 3300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4219506790454687e-05,
+      "loss": 0.7223,
+      "step": 3301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.421612098948375e-05,
+      "loss": 0.7012,
+      "step": 3302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4212734600602569e-05,
+      "loss": 0.8283,
+      "step": 3303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4209347624283352e-05,
+      "loss": 0.8132,
+      "step": 3304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.420596006099839e-05,
+      "loss": 0.9638,
+      "step": 3305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4202571911220063e-05,
+      "loss": 1.0102,
+      "step": 3306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4199183175420816e-05,
+      "loss": 0.7106,
+      "step": 3307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4195793854073195e-05,
+      "loss": 0.6728,
+      "step": 3308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.419240394764981e-05,
+      "loss": 0.7182,
+      "step": 3309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4189013456623371e-05,
+      "loss": 0.6784,
+      "step": 3310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4185622381466653e-05,
+      "loss": 0.9706,
+      "step": 3311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4182230722652518e-05,
+      "loss": 1.16,
+      "step": 3312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4178838480653916e-05,
+      "loss": 0.6484,
+      "step": 3313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4175445655943864e-05,
+      "loss": 0.7743,
+      "step": 3314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.417205224899548e-05,
+      "loss": 0.633,
+      "step": 3315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4168658260281944e-05,
+      "loss": 0.8118,
+      "step": 3316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4165263690276532e-05,
+      "loss": 0.5324,
+      "step": 3317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4161868539452585e-05,
+      "loss": 0.7454,
+      "step": 3318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4158472808283543e-05,
+      "loss": 0.6724,
+      "step": 3319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4155076497242915e-05,
+      "loss": 0.8152,
+      "step": 3320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4151679606804295e-05,
+      "loss": 0.7058,
+      "step": 3321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4148282137441352e-05,
+      "loss": 0.7594,
+      "step": 3322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4144884089627848e-05,
+      "loss": 0.9194,
+      "step": 3323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4141485463837614e-05,
+      "loss": 0.6728,
+      "step": 3324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4138086260544568e-05,
+      "loss": 0.7068,
+      "step": 3325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4134686480222706e-05,
+      "loss": 0.6722,
+      "step": 3326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4131286123346106e-05,
+      "loss": 0.7631,
+      "step": 3327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4127885190388922e-05,
+      "loss": 0.6308,
+      "step": 3328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4124483681825393e-05,
+      "loss": 0.7049,
+      "step": 3329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4121081598129838e-05,
+      "loss": 0.7667,
+      "step": 3330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4117678939776653e-05,
+      "loss": 0.7507,
+      "step": 3331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.411427570724032e-05,
+      "loss": 0.8236,
+      "step": 3332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4110871900995395e-05,
+      "loss": 0.7477,
+      "step": 3333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4107467521516515e-05,
+      "loss": 0.6037,
+      "step": 3334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4104062569278403e-05,
+      "loss": 0.7715,
+      "step": 3335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.410065704475585e-05,
+      "loss": 0.7121,
+      "step": 3336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409725094842374e-05,
+      "loss": 0.6803,
+      "step": 3337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4093844280757028e-05,
+      "loss": 0.9313,
+      "step": 3338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4090437042230751e-05,
+      "loss": 0.6776,
+      "step": 3339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4087029233320026e-05,
+      "loss": 0.6877,
+      "step": 3340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408362085450005e-05,
+      "loss": 0.6659,
+      "step": 3341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4080211906246101e-05,
+      "loss": 0.8007,
+      "step": 3342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4076802389033533e-05,
+      "loss": 0.7057,
+      "step": 3343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4073392303337778e-05,
+      "loss": 1.1025,
+      "step": 3344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4069981649634352e-05,
+      "loss": 0.6853,
+      "step": 3345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406657042839885e-05,
+      "loss": 0.6736,
+      "step": 3346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.406315864010694e-05,
+      "loss": 0.9662,
+      "step": 3347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4059746285234377e-05,
+      "loss": 0.7898,
+      "step": 3348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4056333364256988e-05,
+      "loss": 0.9644,
+      "step": 3349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4052919877650685e-05,
+      "loss": 0.8925,
+      "step": 3350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.404950582589146e-05,
+      "loss": 0.7414,
+      "step": 3351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.404609120945537e-05,
+      "loss": 0.7836,
+      "step": 3352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.404267602881857e-05,
+      "loss": 0.6175,
+      "step": 3353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4039260284457279e-05,
+      "loss": 0.6876,
+      "step": 3354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4035843976847804e-05,
+      "loss": 0.7634,
+      "step": 3355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4032427106466519e-05,
+      "loss": 0.9147,
+      "step": 3356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4029009673789893e-05,
+      "loss": 1.0813,
+      "step": 3357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4025591679294461e-05,
+      "loss": 0.9669,
+      "step": 3358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.402217312345684e-05,
+      "loss": 0.5855,
+      "step": 3359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.401875400675372e-05,
+      "loss": 0.7358,
+      "step": 3360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4015334329661885e-05,
+      "loss": 1.2144,
+      "step": 3361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4011914092658176e-05,
+      "loss": 0.7381,
+      "step": 3362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.400849329621953e-05,
+      "loss": 0.6513,
+      "step": 3363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4005071940822947e-05,
+      "loss": 0.5951,
+      "step": 3364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.400165002694552e-05,
+      "loss": 0.6904,
+      "step": 3365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3998227555064407e-05,
+      "loss": 0.7789,
+      "step": 3366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3994804525656852e-05,
+      "loss": 1.0092,
+      "step": 3367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3991380939200169e-05,
+      "loss": 0.6431,
+      "step": 3368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3987956796171765e-05,
+      "loss": 0.6249,
+      "step": 3369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3984532097049104e-05,
+      "loss": 0.7584,
+      "step": 3370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3981106842309741e-05,
+      "loss": 0.7523,
+      "step": 3371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3977681032431307e-05,
+      "loss": 0.6009,
+      "step": 3372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3974254667891507e-05,
+      "loss": 0.6603,
+      "step": 3373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3970827749168123e-05,
+      "loss": 0.6738,
+      "step": 3374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3967400276739018e-05,
+      "loss": 0.9647,
+      "step": 3375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3963972251082132e-05,
+      "loss": 0.7017,
+      "step": 3376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3960543672675478e-05,
+      "loss": 1.0631,
+      "step": 3377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3957114541997152e-05,
+      "loss": 0.6869,
+      "step": 3378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3953684859525316e-05,
+      "loss": 0.6412,
+      "step": 3379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3950254625738229e-05,
+      "loss": 0.6038,
+      "step": 3380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.39468238411142e-05,
+      "loss": 0.8163,
+      "step": 3381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3943392506131643e-05,
+      "loss": 0.8984,
+      "step": 3382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3939960621269025e-05,
+      "loss": 0.7902,
+      "step": 3383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3936528187004904e-05,
+      "loss": 0.6961,
+      "step": 3384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.393309520381791e-05,
+      "loss": 0.6714,
+      "step": 3385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.392966167218675e-05,
+      "loss": 0.6445,
+      "step": 3386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3926227592590204e-05,
+      "loss": 0.706,
+      "step": 3387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3922792965507139e-05,
+      "loss": 0.795,
+      "step": 3388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3919357791416483e-05,
+      "loss": 0.6084,
+      "step": 3389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3915922070797254e-05,
+      "loss": 0.8139,
+      "step": 3390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3912485804128536e-05,
+      "loss": 0.827,
+      "step": 3391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.39090489918895e-05,
+      "loss": 0.6701,
+      "step": 3392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3905611634559382e-05,
+      "loss": 0.9718,
+      "step": 3393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3902173732617499e-05,
+      "loss": 1.0325,
+      "step": 3394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3898735286543244e-05,
+      "loss": 0.7594,
+      "step": 3395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3895296296816086e-05,
+      "loss": 1.0279,
+      "step": 3396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.389185676391557e-05,
+      "loss": 1.1059,
+      "step": 3397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3888416688321315e-05,
+      "loss": 0.6306,
+      "step": 3398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3884976070513019e-05,
+      "loss": 0.8871,
+      "step": 3399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3881534910970454e-05,
+      "loss": 0.7278,
+      "step": 3400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.387809321017346e-05,
+      "loss": 0.8402,
+      "step": 3401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3874650968601968e-05,
+      "loss": 1.0371,
+      "step": 3402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3871208186735971e-05,
+      "loss": 0.8581,
+      "step": 3403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3867764865055545e-05,
+      "loss": 0.6542,
+      "step": 3404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3864321004040835e-05,
+      "loss": 0.9299,
+      "step": 3405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3860876604172066e-05,
+      "loss": 0.6516,
+      "step": 3406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3857431665929541e-05,
+      "loss": 1.2564,
+      "step": 3407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3853986189793629e-05,
+      "loss": 1.157,
+      "step": 3408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.385054017624478e-05,
+      "loss": 0.642,
+      "step": 3409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3847093625763517e-05,
+      "loss": 0.6031,
+      "step": 3410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3843646538830441e-05,
+      "loss": 0.6388,
+      "step": 3411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3840198915926226e-05,
+      "loss": 0.6277,
+      "step": 3412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3836750757531618e-05,
+      "loss": 0.6085,
+      "step": 3413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3833302064127437e-05,
+      "loss": 0.7166,
+      "step": 3414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3829852836194588e-05,
+      "loss": 0.7662,
+      "step": 3415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3826403074214037e-05,
+      "loss": 0.7409,
+      "step": 3416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3822952778666836e-05,
+      "loss": 0.6267,
+      "step": 3417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3819501950034098e-05,
+      "loss": 0.7386,
+      "step": 3418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3816050588797028e-05,
+      "loss": 0.8175,
+      "step": 3419
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3812598695436887e-05,
+      "loss": 1.0283,
+      "step": 3420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3809146270435023e-05,
+      "loss": 0.6509,
+      "step": 3421
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3805693314272854e-05,
+      "loss": 0.7004,
+      "step": 3422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.380223982743187e-05,
+      "loss": 0.5633,
+      "step": 3423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.379878581039364e-05,
+      "loss": 0.6545,
+      "step": 3424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3795331263639799e-05,
+      "loss": 0.6675,
+      "step": 3425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3791876187652067e-05,
+      "loss": 0.7355,
+      "step": 3426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3788420582912226e-05,
+      "loss": 0.9405,
+      "step": 3427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.378496444990214e-05,
+      "loss": 0.7827,
+      "step": 3428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3781507789103743e-05,
+      "loss": 0.6551,
+      "step": 3429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3778050600999047e-05,
+      "loss": 0.8049,
+      "step": 3430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3774592886070126e-05,
+      "loss": 0.8557,
+      "step": 3431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3771134644799143e-05,
+      "loss": 0.7824,
+      "step": 3432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3767675877668323e-05,
+      "loss": 1.0464,
+      "step": 3433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.376421658515997e-05,
+      "loss": 0.558,
+      "step": 3434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.376075676775646e-05,
+      "loss": 0.6442,
+      "step": 3435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3757296425940243e-05,
+      "loss": 1.0387,
+      "step": 3436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3753835560193835e-05,
+      "loss": 0.6958,
+      "step": 3437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3750374170999838e-05,
+      "loss": 0.6405,
+      "step": 3438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3746912258840914e-05,
+      "loss": 0.7773,
+      "step": 3439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3743449824199809e-05,
+      "loss": 0.6246,
+      "step": 3440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.373998686755933e-05,
+      "loss": 0.7182,
+      "step": 3441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.373652338940237e-05,
+      "loss": 0.9925,
+      "step": 3442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3733059390211882e-05,
+      "loss": 0.7319,
+      "step": 3443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3729594870470906e-05,
+      "loss": 0.6161,
+      "step": 3444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.372612983066254e-05,
+      "loss": 0.6749,
+      "step": 3445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3722664271269965e-05,
+      "loss": 0.6321,
+      "step": 3446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3719198192776424e-05,
+      "loss": 0.7618,
+      "step": 3447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3715731595665247e-05,
+      "loss": 0.6394,
+      "step": 3448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.371226448041982e-05,
+      "loss": 0.7054,
+      "step": 3449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3708796847523617e-05,
+      "loss": 0.6958,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.370532869746017e-05,
+      "loss": 0.5423,
+      "step": 3451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3701860030713095e-05,
+      "loss": 0.7551,
+      "step": 3452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.369839084776607e-05,
+      "loss": 0.6149,
+      "step": 3453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3694921149102851e-05,
+      "loss": 0.615,
+      "step": 3454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3691450935207269e-05,
+      "loss": 0.574,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3687980206563216e-05,
+      "loss": 0.6305,
+      "step": 3456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3684508963654667e-05,
+      "loss": 0.5785,
+      "step": 3457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.368103720696566e-05,
+      "loss": 0.7403,
+      "step": 3458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3677564936980312e-05,
+      "loss": 1.0119,
+      "step": 3459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3674092154182803e-05,
+      "loss": 0.7663,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3670618859057397e-05,
+      "loss": 1.0328,
+      "step": 3461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3667145052088418e-05,
+      "loss": 0.7038,
+      "step": 3462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3663670733760265e-05,
+      "loss": 0.8475,
+      "step": 3463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3660195904557409e-05,
+      "loss": 0.7988,
+      "step": 3464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3656720564964395e-05,
+      "loss": 0.8722,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.365324471546583e-05,
+      "loss": 0.9102,
+      "step": 3466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3649768356546407e-05,
+      "loss": 1.118,
+      "step": 3467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3646291488690872e-05,
+      "loss": 0.7578,
+      "step": 3468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3642814112384054e-05,
+      "loss": 0.9095,
+      "step": 3469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3639336228110857e-05,
+      "loss": 0.7568,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3635857836356241e-05,
+      "loss": 0.576,
+      "step": 3471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3632378937605249e-05,
+      "loss": 0.6538,
+      "step": 3472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3628899532342987e-05,
+      "loss": 0.6241,
+      "step": 3473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3625419621054639e-05,
+      "loss": 0.6586,
+      "step": 3474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3621939204225452e-05,
+      "loss": 0.599,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3618458282340749e-05,
+      "loss": 0.7569,
+      "step": 3476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3614976855885922e-05,
+      "loss": 0.9051,
+      "step": 3477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3611494925346434e-05,
+      "loss": 0.6832,
+      "step": 3478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3608012491207812e-05,
+      "loss": 1.0649,
+      "step": 3479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3604529553955664e-05,
+      "loss": 0.9668,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.360104611407566e-05,
+      "loss": 0.7815,
+      "step": 3481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3597562172053547e-05,
+      "loss": 0.5795,
+      "step": 3482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3594077728375129e-05,
+      "loss": 1.0426,
+      "step": 3483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3590592783526298e-05,
+      "loss": 0.7775,
+      "step": 3484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3587107337993e-05,
+      "loss": 0.8019,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3583621392261262e-05,
+      "loss": 0.7489,
+      "step": 3486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3580134946817171e-05,
+      "loss": 0.6637,
+      "step": 3487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.357664800214689e-05,
+      "loss": 0.6202,
+      "step": 3488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3573160558736657e-05,
+      "loss": 0.7601,
+      "step": 3489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3569672617072767e-05,
+      "loss": 0.5899,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.356618417764159e-05,
+      "loss": 0.5176,
+      "step": 3491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3562695240929568e-05,
+      "loss": 0.6267,
+      "step": 3492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.355920580742321e-05,
+      "loss": 1.0101,
+      "step": 3493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3555715877609092e-05,
+      "loss": 0.4943,
+      "step": 3494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3552225451973869e-05,
+      "loss": 0.6235,
+      "step": 3495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3548734531004246e-05,
+      "loss": 1.002,
+      "step": 3496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3545243115187017e-05,
+      "loss": 0.7085,
+      "step": 3497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3541751205009037e-05,
+      "loss": 0.6833,
+      "step": 3498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3538258800957229e-05,
+      "loss": 0.9101,
+      "step": 3499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3534765903518583e-05,
+      "loss": 1.0052,
+      "step": 3500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3531272513180163e-05,
+      "loss": 0.6875,
+      "step": 3501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3527778630429099e-05,
+      "loss": 0.6731,
+      "step": 3502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.352428425575259e-05,
+      "loss": 0.9059,
+      "step": 3503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3520789389637898e-05,
+      "loss": 0.7849,
+      "step": 3504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.351729403257237e-05,
+      "loss": 0.7615,
+      "step": 3505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.35137981850434e-05,
+      "loss": 0.7327,
+      "step": 3506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3510301847538468e-05,
+      "loss": 0.5621,
+      "step": 3507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.350680502054511e-05,
+      "loss": 0.7097,
+      "step": 3508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.350330770455094e-05,
+      "loss": 0.5978,
+      "step": 3509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3499809900043632e-05,
+      "loss": 0.6234,
+      "step": 3510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3496311607510937e-05,
+      "loss": 0.7812,
+      "step": 3511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.349281282744066e-05,
+      "loss": 0.8741,
+      "step": 3512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.348931356032069e-05,
+      "loss": 0.529,
+      "step": 3513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3485813806638977e-05,
+      "loss": 0.7431,
+      "step": 3514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.348231356688353e-05,
+      "loss": 0.6479,
+      "step": 3515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3478812841542447e-05,
+      "loss": 0.6805,
+      "step": 3516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.347531163110387e-05,
+      "loss": 1.0276,
+      "step": 3517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3471809936056029e-05,
+      "loss": 0.7183,
+      "step": 3518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3468307756887205e-05,
+      "loss": 0.7036,
+      "step": 3519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3464805094085756e-05,
+      "loss": 0.8716,
+      "step": 3520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3461301948140104e-05,
+      "loss": 0.6362,
+      "step": 3521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3457798319538744e-05,
+      "loss": 0.5821,
+      "step": 3522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.345429420877023e-05,
+      "loss": 0.5276,
+      "step": 3523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3450789616323185e-05,
+      "loss": 0.8934,
+      "step": 3524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3447284542686303e-05,
+      "loss": 0.5922,
+      "step": 3525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3443778988348351e-05,
+      "loss": 0.8671,
+      "step": 3526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3440272953798146e-05,
+      "loss": 0.6967,
+      "step": 3527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3436766439524585e-05,
+      "loss": 0.676,
+      "step": 3528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3433259446016627e-05,
+      "loss": 0.6522,
+      "step": 3529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.34297519737633e-05,
+      "loss": 0.6176,
+      "step": 3530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3426244023253698e-05,
+      "loss": 0.7831,
+      "step": 3531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3422735594976981e-05,
+      "loss": 0.6869,
+      "step": 3532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3419226689422375e-05,
+      "loss": 0.7596,
+      "step": 3533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3415717307079177e-05,
+      "loss": 0.7686,
+      "step": 3534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3412207448436746e-05,
+      "loss": 0.9625,
+      "step": 3535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3408697113984506e-05,
+      "loss": 0.684,
+      "step": 3536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3405186304211954e-05,
+      "loss": 0.896,
+      "step": 3537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3401675019608646e-05,
+      "loss": 0.6044,
+      "step": 3538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.339816326066421e-05,
+      "loss": 1.0833,
+      "step": 3539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3394651027868333e-05,
+      "loss": 0.6379,
+      "step": 3540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.339113832171078e-05,
+      "loss": 1.0442,
+      "step": 3541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3387625142681367e-05,
+      "loss": 0.8458,
+      "step": 3542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3384111491269986e-05,
+      "loss": 0.7532,
+      "step": 3543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3380597367966596e-05,
+      "loss": 0.637,
+      "step": 3544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.337708277326122e-05,
+      "loss": 0.7298,
+      "step": 3545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3373567707643935e-05,
+      "loss": 0.7976,
+      "step": 3546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3370052171604904e-05,
+      "loss": 0.6619,
+      "step": 3547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3366536165634338e-05,
+      "loss": 0.9341,
+      "step": 3548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3363019690222528e-05,
+      "loss": 0.8245,
+      "step": 3549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3359502745859815e-05,
+      "loss": 0.6967,
+      "step": 3550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.335598533303662e-05,
+      "loss": 0.5927,
+      "step": 3551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3352467452243421e-05,
+      "loss": 0.6546,
+      "step": 3552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.334894910397076e-05,
+      "loss": 0.6307,
+      "step": 3553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3345430288709254e-05,
+      "loss": 0.6036,
+      "step": 3554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3341911006949573e-05,
+      "loss": 0.6406,
+      "step": 3555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3338391259182461e-05,
+      "loss": 0.7082,
+      "step": 3556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3334871045898722e-05,
+      "loss": 0.7248,
+      "step": 3557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3331350367589227e-05,
+      "loss": 1.0125,
+      "step": 3558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3327829224744911e-05,
+      "loss": 0.6307,
+      "step": 3559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3324307617856775e-05,
+      "loss": 0.7491,
+      "step": 3560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.332078554741588e-05,
+      "loss": 0.7612,
+      "step": 3561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3317263013913362e-05,
+      "loss": 0.7369,
+      "step": 3562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3313740017840412e-05,
+      "loss": 0.6231,
+      "step": 3563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3310216559688288e-05,
+      "loss": 0.9867,
+      "step": 3564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3306692639948309e-05,
+      "loss": 0.7318,
+      "step": 3565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3303168259111872e-05,
+      "loss": 0.5616,
+      "step": 3566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.329964341767042e-05,
+      "loss": 0.7559,
+      "step": 3567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3296118116115473e-05,
+      "loss": 0.8014,
+      "step": 3568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.329259235493861e-05,
+      "loss": 0.6588,
+      "step": 3569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3289066134631477e-05,
+      "loss": 1.0127,
+      "step": 3570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3285539455685777e-05,
+      "loss": 1.1867,
+      "step": 3571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3282012318593287e-05,
+      "loss": 0.7683,
+      "step": 3572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3278484723845842e-05,
+      "loss": 0.7999,
+      "step": 3573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3274956671935342e-05,
+      "loss": 0.8563,
+      "step": 3574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3271428163353748e-05,
+      "loss": 0.5286,
+      "step": 3575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3267899198593091e-05,
+      "loss": 0.7112,
+      "step": 3576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3264369778145458e-05,
+      "loss": 0.7018,
+      "step": 3577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3260839902503007e-05,
+      "loss": 0.642,
+      "step": 3578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3257309572157953e-05,
+      "loss": 0.8035,
+      "step": 3579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3253778787602578e-05,
+      "loss": 0.6291,
+      "step": 3580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.325024754932923e-05,
+      "loss": 1.0577,
+      "step": 3581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.324671585783031e-05,
+      "loss": 0.6302,
+      "step": 3582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3243183713598296e-05,
+      "loss": 0.6501,
+      "step": 3583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3239651117125716e-05,
+      "loss": 1.0872,
+      "step": 3584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3236118068905174e-05,
+      "loss": 0.6906,
+      "step": 3585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3232584569429326e-05,
+      "loss": 0.8481,
+      "step": 3586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3229050619190895e-05,
+      "loss": 0.6197,
+      "step": 3587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3225516218682666e-05,
+      "loss": 0.9615,
+      "step": 3588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3221981368397491e-05,
+      "loss": 0.7896,
+      "step": 3589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3218446068828279e-05,
+      "loss": 0.6116,
+      "step": 3590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3214910320468009e-05,
+      "loss": 0.7593,
+      "step": 3591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.321137412380971e-05,
+      "loss": 0.6375,
+      "step": 3592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.320783747934649e-05,
+      "loss": 0.7044,
+      "step": 3593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3204300387571506e-05,
+      "loss": 0.6736,
+      "step": 3594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3200762848977979e-05,
+      "loss": 0.6662,
+      "step": 3595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.31972248640592e-05,
+      "loss": 0.7902,
+      "step": 3596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3193686433308517e-05,
+      "loss": 0.7884,
+      "step": 3597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.319014755721934e-05,
+      "loss": 0.6512,
+      "step": 3598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3186608236285138e-05,
+      "loss": 0.6519,
+      "step": 3599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3183068470999457e-05,
+      "loss": 0.6321,
+      "step": 3600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3179528261855883e-05,
+      "loss": 0.9471,
+      "step": 3601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.317598760934808e-05,
+      "loss": 0.6941,
+      "step": 3602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3172446513969767e-05,
+      "loss": 1.0101,
+      "step": 3603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.316890497621473e-05,
+      "loss": 0.894,
+      "step": 3604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3165362996576804e-05,
+      "loss": 1.002,
+      "step": 3605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3161820575549904e-05,
+      "loss": 0.656,
+      "step": 3606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3158277713627992e-05,
+      "loss": 0.6832,
+      "step": 3607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3154734411305099e-05,
+      "loss": 0.7076,
+      "step": 3608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3151190669075313e-05,
+      "loss": 0.655,
+      "step": 3609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.314764648743279e-05,
+      "loss": 0.83,
+      "step": 3610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3144101866871739e-05,
+      "loss": 0.6008,
+      "step": 3611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3140556807886436e-05,
+      "loss": 1.1706,
+      "step": 3612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3137011310971213e-05,
+      "loss": 0.6131,
+      "step": 3613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3133465376620469e-05,
+      "loss": 0.5798,
+      "step": 3614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3129919005328657e-05,
+      "loss": 0.7877,
+      "step": 3615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3126372197590303e-05,
+      "loss": 0.6722,
+      "step": 3616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3122824953899975e-05,
+      "loss": 0.6601,
+      "step": 3617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3119277274752325e-05,
+      "loss": 0.6637,
+      "step": 3618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3115729160642041e-05,
+      "loss": 0.5563,
+      "step": 3619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3112180612063897e-05,
+      "loss": 0.8673,
+      "step": 3620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3108631629512703e-05,
+      "loss": 0.6411,
+      "step": 3621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105082213483348e-05,
+      "loss": 0.7541,
+      "step": 3622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101532364470773e-05,
+      "loss": 0.8043,
+      "step": 3623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3097982082969982e-05,
+      "loss": 1.0664,
+      "step": 3624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3094431369476035e-05,
+      "loss": 0.9038,
+      "step": 3625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3090880224484059e-05,
+      "loss": 0.9395,
+      "step": 3626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3087328648489237e-05,
+      "loss": 0.7585,
+      "step": 3627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3083776641986812e-05,
+      "loss": 0.6829,
+      "step": 3628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3080224205472088e-05,
+      "loss": 1.0526,
+      "step": 3629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3076671339440427e-05,
+      "loss": 0.7382,
+      "step": 3630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3073118044387258e-05,
+      "loss": 0.6072,
+      "step": 3631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306956432080806e-05,
+      "loss": 0.6966,
+      "step": 3632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066010169198378e-05,
+      "loss": 0.6671,
+      "step": 3633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3062455590053811e-05,
+      "loss": 0.6479,
+      "step": 3634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3058900583870027e-05,
+      "loss": 0.9591,
+      "step": 3635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3055345151142746e-05,
+      "loss": 0.7669,
+      "step": 3636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3051789292367754e-05,
+      "loss": 0.6598,
+      "step": 3637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3048233008040882e-05,
+      "loss": 0.9423,
+      "step": 3638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.304467629865804e-05,
+      "loss": 0.7293,
+      "step": 3639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3041119164715184e-05,
+      "loss": 0.7397,
+      "step": 3640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3037561606708333e-05,
+      "loss": 0.8479,
+      "step": 3641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3034003625133562e-05,
+      "loss": 0.7872,
+      "step": 3642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030445220487016e-05,
+      "loss": 0.918,
+      "step": 3643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3026886393264882e-05,
+      "loss": 0.7986,
+      "step": 3644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3023327143963415e-05,
+      "loss": 0.6989,
+      "step": 3645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301976747307894e-05,
+      "loss": 0.7565,
+      "step": 3646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3016207381107818e-05,
+      "loss": 0.6318,
+      "step": 3647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012646868546489e-05,
+      "loss": 0.7299,
+      "step": 3648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3009085935891434e-05,
+      "loss": 0.7374,
+      "step": 3649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300552458363921e-05,
+      "loss": 0.6521,
+      "step": 3650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3001962812286415e-05,
+      "loss": 0.7294,
+      "step": 3651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2998400622329724e-05,
+      "loss": 0.6046,
+      "step": 3652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2994838014265851e-05,
+      "loss": 0.872,
+      "step": 3653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2991274988591591e-05,
+      "loss": 0.7237,
+      "step": 3654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987711545803773e-05,
+      "loss": 0.5975,
+      "step": 3655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984147686399302e-05,
+      "loss": 0.615,
+      "step": 3656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2980583410875131e-05,
+      "loss": 0.8854,
+      "step": 3657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2977018719728278e-05,
+      "loss": 0.5956,
+      "step": 3658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2973453613455811e-05,
+      "loss": 0.6122,
+      "step": 3659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2969888092554865e-05,
+      "loss": 0.6682,
+      "step": 3660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2966322157522626e-05,
+      "loss": 0.6318,
+      "step": 3661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2962755808856341e-05,
+      "loss": 0.6128,
+      "step": 3662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2959189047053315e-05,
+      "loss": 0.6401,
+      "step": 3663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2955621872610905e-05,
+      "loss": 1.0714,
+      "step": 3664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2952054286026536e-05,
+      "loss": 0.7062,
+      "step": 3665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2948486287797679e-05,
+      "loss": 0.6119,
+      "step": 3666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2944917878421873e-05,
+      "loss": 0.954,
+      "step": 3667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2941349058396701e-05,
+      "loss": 0.7674,
+      "step": 3668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2937779828219823e-05,
+      "loss": 0.6175,
+      "step": 3669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2934210188388935e-05,
+      "loss": 0.7532,
+      "step": 3670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2930640139401801e-05,
+      "loss": 0.6178,
+      "step": 3671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2927069681756244e-05,
+      "loss": 0.6732,
+      "step": 3672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2923498815950142e-05,
+      "loss": 0.7266,
+      "step": 3673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2919927542481423e-05,
+      "loss": 0.6821,
+      "step": 3674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2916355861848083e-05,
+      "loss": 0.8074,
+      "step": 3675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2912783774548163e-05,
+      "loss": 0.6838,
+      "step": 3676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2909211281079776e-05,
+      "loss": 0.8864,
+      "step": 3677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2905638381941074e-05,
+      "loss": 0.9613,
+      "step": 3678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.290206507763028e-05,
+      "loss": 0.7406,
+      "step": 3679
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2898491368645663e-05,
+      "loss": 0.7953,
+      "step": 3680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2894917255485558e-05,
+      "loss": 0.8272,
+      "step": 3681
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2891342738648347e-05,
+      "loss": 0.7328,
+      "step": 3682
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2887767818632477e-05,
+      "loss": 0.7258,
+      "step": 3683
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2884192495936443e-05,
+      "loss": 0.7304,
+      "step": 3684
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2880616771058802e-05,
+      "loss": 0.7472,
+      "step": 3685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2877040644498167e-05,
+      "loss": 0.7423,
+      "step": 3686
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2873464116753205e-05,
+      "loss": 1.025,
+      "step": 3687
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2869887188322632e-05,
+      "loss": 1.0299,
+      "step": 3688
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2866309859705236e-05,
+      "loss": 0.7669,
+      "step": 3689
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2862732131399849e-05,
+      "loss": 0.616,
+      "step": 3690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.285915400390536e-05,
+      "loss": 0.6556,
+      "step": 3691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.285557547772072e-05,
+      "loss": 0.8781,
+      "step": 3692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2851996553344922e-05,
+      "loss": 0.654,
+      "step": 3693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2848417231277032e-05,
+      "loss": 0.7252,
+      "step": 3694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2844837512016159e-05,
+      "loss": 0.5574,
+      "step": 3695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2841257396061474e-05,
+      "loss": 0.691,
+      "step": 3696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2837676883912193e-05,
+      "loss": 0.7275,
+      "step": 3697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2834095976067606e-05,
+      "loss": 0.5927,
+      "step": 3698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2830514673027035e-05,
+      "loss": 0.5912,
+      "step": 3699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2826932975289878e-05,
+      "loss": 0.716,
+      "step": 3700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2823350883355577e-05,
+      "loss": 0.6452,
+      "step": 3701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.281976839772363e-05,
+      "loss": 1.1602,
+      "step": 3702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2816185518893591e-05,
+      "loss": 0.5417,
+      "step": 3703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2812602247365071e-05,
+      "loss": 0.7137,
+      "step": 3704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2809018583637728e-05,
+      "loss": 1.0033,
+      "step": 3705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2805434528211287e-05,
+      "loss": 0.5591,
+      "step": 3706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2801850081585516e-05,
+      "loss": 0.88,
+      "step": 3707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2798265244260246e-05,
+      "loss": 0.8387,
+      "step": 3708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2794680016735355e-05,
+      "loss": 0.7452,
+      "step": 3709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2791094399510781e-05,
+      "loss": 0.8317,
+      "step": 3710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2787508393086518e-05,
+      "loss": 0.969,
+      "step": 3711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2783921997962604e-05,
+      "loss": 0.6023,
+      "step": 3712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2780335214639146e-05,
+      "loss": 0.5877,
+      "step": 3713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2776748043616292e-05,
+      "loss": 0.6277,
+      "step": 3714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2773160485394252e-05,
+      "loss": 0.6582,
+      "step": 3715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2769572540473283e-05,
+      "loss": 0.6962,
+      "step": 3716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2765984209353706e-05,
+      "loss": 0.7714,
+      "step": 3717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2762395492535884e-05,
+      "loss": 0.6703,
+      "step": 3718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2758806390520246e-05,
+      "loss": 0.6397,
+      "step": 3719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2755216903807263e-05,
+      "loss": 0.6304,
+      "step": 3720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2751627032897473e-05,
+      "loss": 0.8188,
+      "step": 3721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2748036778291452e-05,
+      "loss": 0.8332,
+      "step": 3722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.274444614048984e-05,
+      "loss": 0.9351,
+      "step": 3723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2740855119993329e-05,
+      "loss": 0.7309,
+      "step": 3724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2737263717302661e-05,
+      "loss": 0.8462,
+      "step": 3725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2733671932918634e-05,
+      "loss": 0.6355,
+      "step": 3726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.27300797673421e-05,
+      "loss": 0.751,
+      "step": 3727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.272648722107396e-05,
+      "loss": 0.6695,
+      "step": 3728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2722894294615178e-05,
+      "loss": 0.6851,
+      "step": 3729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2719300988466752e-05,
+      "loss": 0.8725,
+      "step": 3730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2715707303129756e-05,
+      "loss": 0.9231,
+      "step": 3731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2712113239105299e-05,
+      "loss": 0.6933,
+      "step": 3732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2708518796894553e-05,
+      "loss": 0.6732,
+      "step": 3733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2704923976998736e-05,
+      "loss": 0.5851,
+      "step": 3734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2701328779919125e-05,
+      "loss": 0.6938,
+      "step": 3735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2697733206157046e-05,
+      "loss": 0.8062,
+      "step": 3736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2694137256213874e-05,
+      "loss": 0.7952,
+      "step": 3737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2690540930591046e-05,
+      "loss": 0.7115,
+      "step": 3738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2686944229790044e-05,
+      "loss": 0.6607,
+      "step": 3739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2683347154312404e-05,
+      "loss": 0.6734,
+      "step": 3740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2679749704659711e-05,
+      "loss": 0.7332,
+      "step": 3741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2676151881333614e-05,
+      "loss": 0.6689,
+      "step": 3742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2672553684835798e-05,
+      "loss": 1.1618,
+      "step": 3743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2668955115668011e-05,
+      "loss": 0.6755,
+      "step": 3744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2665356174332044e-05,
+      "loss": 0.5402,
+      "step": 3745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2661756861329756e-05,
+      "loss": 0.7064,
+      "step": 3746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2658157177163042e-05,
+      "loss": 0.6234,
+      "step": 3747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2654557122333855e-05,
+      "loss": 0.9091,
+      "step": 3748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2650956697344197e-05,
+      "loss": 0.6215,
+      "step": 3749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2647355902696123e-05,
+      "loss": 0.7095,
+      "step": 3750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2643754738891744e-05,
+      "loss": 0.6454,
+      "step": 3751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2640153206433216e-05,
+      "loss": 0.7338,
+      "step": 3752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2636551305822748e-05,
+      "loss": 0.6405,
+      "step": 3753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2632949037562606e-05,
+      "loss": 0.6803,
+      "step": 3754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2629346402155094e-05,
+      "loss": 0.7829,
+      "step": 3755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2625743400102582e-05,
+      "loss": 0.6067,
+      "step": 3756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2622140031907484e-05,
+      "loss": 0.9209,
+      "step": 3757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2618536298072266e-05,
+      "loss": 0.6978,
+      "step": 3758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2614932199099446e-05,
+      "loss": 0.7832,
+      "step": 3759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2611327735491588e-05,
+      "loss": 0.6281,
+      "step": 3760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2607722907751315e-05,
+      "loss": 0.663,
+      "step": 3761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.260411771638129e-05,
+      "loss": 1.0389,
+      "step": 3762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2600512161884243e-05,
+      "loss": 0.7555,
+      "step": 3763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2596906244762933e-05,
+      "loss": 0.8344,
+      "step": 3764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2593299965520191e-05,
+      "loss": 0.7157,
+      "step": 3765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2589693324658886e-05,
+      "loss": 0.717,
+      "step": 3766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.258608632268194e-05,
+      "loss": 0.9615,
+      "step": 3767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2582478960092322e-05,
+      "loss": 1.0978,
+      "step": 3768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2578871237393062e-05,
+      "loss": 0.9774,
+      "step": 3769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2575263155087227e-05,
+      "loss": 0.7786,
+      "step": 3770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2571654713677943e-05,
+      "loss": 0.5944,
+      "step": 3771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2568045913668382e-05,
+      "loss": 0.6071,
+      "step": 3772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.256443675556177e-05,
+      "loss": 0.8087,
+      "step": 3773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.256082723986138e-05,
+      "loss": 0.9168,
+      "step": 3774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.255721736707053e-05,
+      "loss": 0.6797,
+      "step": 3775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2553607137692599e-05,
+      "loss": 0.6681,
+      "step": 3776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2549996552231005e-05,
+      "loss": 1.2996,
+      "step": 3777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2546385611189227e-05,
+      "loss": 0.774,
+      "step": 3778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2542774315070778e-05,
+      "loss": 0.7191,
+      "step": 3779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2539162664379236e-05,
+      "loss": 0.6543,
+      "step": 3780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253555065961822e-05,
+      "loss": 0.5862,
+      "step": 3781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2531938301291398e-05,
+      "loss": 0.8506,
+      "step": 3782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2528325589902491e-05,
+      "loss": 0.5602,
+      "step": 3783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.252471252595527e-05,
+      "loss": 0.7356,
+      "step": 3784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.252109910995355e-05,
+      "loss": 0.6851,
+      "step": 3785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2517485342401201e-05,
+      "loss": 0.6636,
+      "step": 3786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2513871223802133e-05,
+      "loss": 1.0658,
+      "step": 3787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2510256754660322e-05,
+      "loss": 0.8341,
+      "step": 3788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.250664193547977e-05,
+      "loss": 0.5837,
+      "step": 3789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2503026766764547e-05,
+      "loss": 0.6027,
+      "step": 3790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2499411249018758e-05,
+      "loss": 1.088,
+      "step": 3791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2495795382746572e-05,
+      "loss": 0.8271,
+      "step": 3792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2492179168452191e-05,
+      "loss": 0.6982,
+      "step": 3793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2488562606639876e-05,
+      "loss": 0.7056,
+      "step": 3794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2484945697813929e-05,
+      "loss": 1.1301,
+      "step": 3795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2481328442478706e-05,
+      "loss": 1.0249,
+      "step": 3796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2477710841138609e-05,
+      "loss": 0.7559,
+      "step": 3797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2474092894298092e-05,
+      "loss": 0.8557,
+      "step": 3798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2470474602461646e-05,
+      "loss": 0.8357,
+      "step": 3799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2466855966133826e-05,
+      "loss": 0.747,
+      "step": 3800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.246323698581922e-05,
+      "loss": 0.9708,
+      "step": 3801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2459617662022476e-05,
+      "loss": 0.702,
+      "step": 3802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2455997995248286e-05,
+      "loss": 0.7238,
+      "step": 3803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2452377986001382e-05,
+      "loss": 0.5641,
+      "step": 3804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2448757634786556e-05,
+      "loss": 1.1476,
+      "step": 3805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.244513694210864e-05,
+      "loss": 0.6681,
+      "step": 3806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.244151590847252e-05,
+      "loss": 0.7865,
+      "step": 3807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2437894534383115e-05,
+      "loss": 0.7075,
+      "step": 3808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2434272820345411e-05,
+      "loss": 0.6289,
+      "step": 3809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2430650766864427e-05,
+      "loss": 0.5434,
+      "step": 3810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.242702837444524e-05,
+      "loss": 0.5948,
+      "step": 3811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2423405643592963e-05,
+      "loss": 0.558,
+      "step": 3812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2419782574812767e-05,
+      "loss": 0.939,
+      "step": 3813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2416159168609861e-05,
+      "loss": 0.6656,
+      "step": 3814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2412535425489504e-05,
+      "loss": 0.5463,
+      "step": 3815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2408911345957008e-05,
+      "loss": 0.5869,
+      "step": 3816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2405286930517724e-05,
+      "loss": 0.7505,
+      "step": 3817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2401662179677052e-05,
+      "loss": 0.8044,
+      "step": 3818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.239803709394044e-05,
+      "loss": 0.6457,
+      "step": 3819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2394411673813386e-05,
+      "loss": 0.5881,
+      "step": 3820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2390785919801426e-05,
+      "loss": 0.8579,
+      "step": 3821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2387159832410152e-05,
+      "loss": 1.0191,
+      "step": 3822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2383533412145193e-05,
+      "loss": 1.0124,
+      "step": 3823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2379906659512233e-05,
+      "loss": 0.8494,
+      "step": 3824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2376279575016996e-05,
+      "loss": 0.7687,
+      "step": 3825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2372652159165258e-05,
+      "loss": 0.6569,
+      "step": 3826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2369024412462833e-05,
+      "loss": 0.5596,
+      "step": 3827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2365396335415593e-05,
+      "loss": 0.7228,
+      "step": 3828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2361767928529445e-05,
+      "loss": 0.5194,
+      "step": 3829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.235813919231035e-05,
+      "loss": 0.7153,
+      "step": 3830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2354510127264307e-05,
+      "loss": 0.7473,
+      "step": 3831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.235088073389737e-05,
+      "loss": 1.0499,
+      "step": 3832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2347251012715629e-05,
+      "loss": 0.9189,
+      "step": 3833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2343620964225227e-05,
+      "loss": 0.6914,
+      "step": 3834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2339990588932348e-05,
+      "loss": 0.9504,
+      "step": 3835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2336359887343232e-05,
+      "loss": 0.5349,
+      "step": 3836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2332728859964145e-05,
+      "loss": 0.625,
+      "step": 3837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2329097507301416e-05,
+      "loss": 0.7082,
+      "step": 3838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2325465829861414e-05,
+      "loss": 0.7789,
+      "step": 3839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2321833828150551e-05,
+      "loss": 0.6308,
+      "step": 3840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2318201502675285e-05,
+      "loss": 0.7673,
+      "step": 3841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2314568853942119e-05,
+      "loss": 0.9544,
+      "step": 3842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2310935882457607e-05,
+      "loss": 0.7152,
+      "step": 3843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2307302588728334e-05,
+      "loss": 1.1638,
+      "step": 3844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2303668973260948e-05,
+      "loss": 0.7056,
+      "step": 3845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2300035036562126e-05,
+      "loss": 0.6253,
+      "step": 3846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2296400779138598e-05,
+      "loss": 0.5823,
+      "step": 3847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.229276620149714e-05,
+      "loss": 0.7116,
+      "step": 3848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.228913130414457e-05,
+      "loss": 1.011,
+      "step": 3849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2285496087587745e-05,
+      "loss": 0.6586,
+      "step": 3850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.228186055233358e-05,
+      "loss": 0.6546,
+      "step": 3851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2278224698889019e-05,
+      "loss": 0.7128,
+      "step": 3852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2274588527761062e-05,
+      "loss": 0.6812,
+      "step": 3853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2270952039456745e-05,
+      "loss": 0.7552,
+      "step": 3854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.226731523448316e-05,
+      "loss": 0.6772,
+      "step": 3855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2263678113347424e-05,
+      "loss": 0.7358,
+      "step": 3856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2260040676556722e-05,
+      "loss": 0.5496,
+      "step": 3857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.225640292461826e-05,
+      "loss": 0.8439,
+      "step": 3858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2252764858039307e-05,
+      "loss": 0.7434,
+      "step": 3859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2249126477327163e-05,
+      "loss": 0.7181,
+      "step": 3860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2245487782989177e-05,
+      "loss": 0.8945,
+      "step": 3861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2241848775532736e-05,
+      "loss": 0.6874,
+      "step": 3862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2238209455465286e-05,
+      "loss": 0.721,
+      "step": 3863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2234569823294295e-05,
+      "loss": 0.6088,
+      "step": 3864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2230929879527294e-05,
+      "loss": 0.6856,
+      "step": 3865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2227289624671847e-05,
+      "loss": 0.8395,
+      "step": 3866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.222364905923556e-05,
+      "loss": 0.7207,
+      "step": 3867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2220008183726093e-05,
+      "loss": 0.6352,
+      "step": 3868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2216366998651132e-05,
+      "loss": 0.8462,
+      "step": 3869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2212725504518428e-05,
+      "loss": 0.6902,
+      "step": 3870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2209083701835752e-05,
+      "loss": 0.8255,
+      "step": 3871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2205441591110937e-05,
+      "loss": 0.885,
+      "step": 3872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2201799172851846e-05,
+      "loss": 0.6588,
+      "step": 3873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2198156447566394e-05,
+      "loss": 0.5931,
+      "step": 3874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2194513415762533e-05,
+      "loss": 0.7955,
+      "step": 3875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2190870077948263e-05,
+      "loss": 0.7821,
+      "step": 3876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.218722643463162e-05,
+      "loss": 0.603,
+      "step": 3877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2183582486320689e-05,
+      "loss": 0.5977,
+      "step": 3878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.217993823352359e-05,
+      "loss": 0.5537,
+      "step": 3879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2176293676748494e-05,
+      "loss": 0.6195,
+      "step": 3880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2172648816503607e-05,
+      "loss": 0.6267,
+      "step": 3881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2169003653297184e-05,
+      "loss": 0.5895,
+      "step": 3882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2165358187637517e-05,
+      "loss": 0.7515,
+      "step": 3883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2161712420032943e-05,
+      "loss": 0.6452,
+      "step": 3884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2158066350991842e-05,
+      "loss": 0.9389,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.215441998102263e-05,
+      "loss": 1.0052,
+      "step": 3886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2150773310633772e-05,
+      "loss": 0.5739,
+      "step": 3887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2147126340333773e-05,
+      "loss": 1.0941,
+      "step": 3888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2143479070631176e-05,
+      "loss": 0.9767,
+      "step": 3889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2139831502034571e-05,
+      "loss": 0.686,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2136183635052587e-05,
+      "loss": 0.6393,
+      "step": 3891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2132535470193893e-05,
+      "loss": 0.966,
+      "step": 3892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2128887007967203e-05,
+      "loss": 0.6308,
+      "step": 3893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2125238248881271e-05,
+      "loss": 0.7931,
+      "step": 3894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2121589193444894e-05,
+      "loss": 0.6596,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2117939842166905e-05,
+      "loss": 0.7682,
+      "step": 3896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2114290195556188e-05,
+      "loss": 0.6961,
+      "step": 3897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2110640254121655e-05,
+      "loss": 0.7956,
+      "step": 3898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.210699001837227e-05,
+      "loss": 0.7451,
+      "step": 3899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2103339488817033e-05,
+      "loss": 0.654,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.209968866596499e-05,
+      "loss": 0.6571,
+      "step": 3901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2096037550325217e-05,
+      "loss": 0.6546,
+      "step": 3902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2092386142406845e-05,
+      "loss": 0.6275,
+      "step": 3903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2088734442719034e-05,
+      "loss": 0.7182,
+      "step": 3904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2085082451770994e-05,
+      "loss": 0.6075,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2081430170071964e-05,
+      "loss": 0.7108,
+      "step": 3906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.207777759813124e-05,
+      "loss": 0.6972,
+      "step": 3907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.207412473645814e-05,
+      "loss": 1.003,
+      "step": 3908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2070471585562038e-05,
+      "loss": 0.8843,
+      "step": 3909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2066818145952337e-05,
+      "loss": 0.809,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2063164418138487e-05,
+      "loss": 0.9609,
+      "step": 3911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2059510402629979e-05,
+      "loss": 0.5671,
+      "step": 3912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2055856099936336e-05,
+      "loss": 0.583,
+      "step": 3913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2052201510567133e-05,
+      "loss": 0.8041,
+      "step": 3914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.204854663503197e-05,
+      "loss": 1.0248,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2044891473840503e-05,
+      "loss": 0.7185,
+      "step": 3916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2041236027502414e-05,
+      "loss": 0.9487,
+      "step": 3917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2037580296527436e-05,
+      "loss": 0.7942,
+      "step": 3918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2033924281425334e-05,
+      "loss": 0.7693,
+      "step": 3919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2030267982705915e-05,
+      "loss": 0.6587,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2026611400879025e-05,
+      "loss": 0.6577,
+      "step": 3921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2022954536454554e-05,
+      "loss": 0.6305,
+      "step": 3922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.201929738994242e-05,
+      "loss": 0.8279,
+      "step": 3923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.20156399618526e-05,
+      "loss": 1.0171,
+      "step": 3924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2011982252695087e-05,
+      "loss": 0.8272,
+      "step": 3925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.200832426297993e-05,
+      "loss": 0.7485,
+      "step": 3926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.200466599321721e-05,
+      "loss": 0.7979,
+      "step": 3927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.200100744391705e-05,
+      "loss": 0.6929,
+      "step": 3928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1997348615589606e-05,
+      "loss": 0.6504,
+      "step": 3929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1993689508745081e-05,
+      "loss": 0.721,
+      "step": 3930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1990030123893716e-05,
+      "loss": 0.6298,
+      "step": 3931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1986370461545782e-05,
+      "loss": 0.9234,
+      "step": 3932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1982710522211601e-05,
+      "loss": 0.7246,
+      "step": 3933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1979050306401521e-05,
+      "loss": 0.6315,
+      "step": 3934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.197538981462594e-05,
+      "loss": 0.7162,
+      "step": 3935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1971729047395287e-05,
+      "loss": 0.584,
+      "step": 3936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1968068005220033e-05,
+      "loss": 0.5305,
+      "step": 3937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1964406688610684e-05,
+      "loss": 0.6735,
+      "step": 3938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1960745098077786e-05,
+      "loss": 0.7285,
+      "step": 3939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1957083234131927e-05,
+      "loss": 0.8015,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1953421097283734e-05,
+      "loss": 0.6563,
+      "step": 3941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1949758688043856e-05,
+      "loss": 0.735,
+      "step": 3942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1946096006923e-05,
+      "loss": 0.6482,
+      "step": 3943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.19424330544319e-05,
+      "loss": 0.6853,
+      "step": 3944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1938769831081333e-05,
+      "loss": 0.8311,
+      "step": 3945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1935106337382107e-05,
+      "loss": 0.9384,
+      "step": 3946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1931442573845078e-05,
+      "loss": 0.8115,
+      "step": 3947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1927778540981127e-05,
+      "loss": 0.5904,
+      "step": 3948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1924114239301182e-05,
+      "loss": 1.0133,
+      "step": 3949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.192044966931621e-05,
+      "loss": 0.5933,
+      "step": 3950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1916784831537203e-05,
+      "loss": 0.7755,
+      "step": 3951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1913119726475207e-05,
+      "loss": 0.9669,
+      "step": 3952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1909454354641288e-05,
+      "loss": 1.1038,
+      "step": 3953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1905788716546564e-05,
+      "loss": 0.6224,
+      "step": 3954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1902122812702181e-05,
+      "loss": 0.7207,
+      "step": 3955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.189845664361933e-05,
+      "loss": 0.612,
+      "step": 3956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1894790209809223e-05,
+      "loss": 0.606,
+      "step": 3957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1891123511783134e-05,
+      "loss": 0.685,
+      "step": 3958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1887456550052353e-05,
+      "loss": 0.8067,
+      "step": 3959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1883789325128213e-05,
+      "loss": 0.8996,
+      "step": 3960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1880121837522083e-05,
+      "loss": 0.9523,
+      "step": 3961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1876454087745374e-05,
+      "loss": 0.7408,
+      "step": 3962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1872786076309528e-05,
+      "loss": 0.5671,
+      "step": 3963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1869117803726027e-05,
+      "loss": 0.8052,
+      "step": 3964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.186544927050638e-05,
+      "loss": 0.5925,
+      "step": 3965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1861780477162146e-05,
+      "loss": 0.6165,
+      "step": 3966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1858111424204912e-05,
+      "loss": 0.7031,
+      "step": 3967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.185444211214631e-05,
+      "loss": 0.659,
+      "step": 3968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1850772541497987e-05,
+      "loss": 0.6001,
+      "step": 3969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1847102712771656e-05,
+      "loss": 0.8126,
+      "step": 3970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1843432626479039e-05,
+      "loss": 0.5918,
+      "step": 3971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1839762283131912e-05,
+      "loss": 0.6896,
+      "step": 3972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1836091683242075e-05,
+      "loss": 0.6847,
+      "step": 3973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1832420827321374e-05,
+      "loss": 0.715,
+      "step": 3974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1828749715881682e-05,
+      "loss": 1.0068,
+      "step": 3975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.182507834943491e-05,
+      "loss": 0.7728,
+      "step": 3976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1821406728493012e-05,
+      "loss": 0.7602,
+      "step": 3977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1817734853567966e-05,
+      "loss": 1.0091,
+      "step": 3978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1814062725171795e-05,
+      "loss": 0.6412,
+      "step": 3979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1810390343816547e-05,
+      "loss": 0.6473,
+      "step": 3980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1806717710014314e-05,
+      "loss": 0.6274,
+      "step": 3981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1803044824277223e-05,
+      "loss": 0.7426,
+      "step": 3982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1799371687117433e-05,
+      "loss": 0.6294,
+      "step": 3983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1795698299047136e-05,
+      "loss": 0.6342,
+      "step": 3984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1792024660578563e-05,
+      "loss": 0.6286,
+      "step": 3985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.178835077222398e-05,
+      "loss": 0.5865,
+      "step": 3986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1784676634495686e-05,
+      "loss": 0.6188,
+      "step": 3987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1781002247906013e-05,
+      "loss": 0.7964,
+      "step": 3988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1777327612967332e-05,
+      "loss": 1.1617,
+      "step": 3989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1773652730192046e-05,
+      "loss": 0.6747,
+      "step": 3990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1769977600092597e-05,
+      "loss": 0.98,
+      "step": 3991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1766302223181452e-05,
+      "loss": 0.6709,
+      "step": 3992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1762626599971118e-05,
+      "loss": 0.8334,
+      "step": 3993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1758950730974141e-05,
+      "loss": 0.9283,
+      "step": 3994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1755274616703093e-05,
+      "loss": 0.6218,
+      "step": 3995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1751598257670586e-05,
+      "loss": 0.7648,
+      "step": 3996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.174792165438926e-05,
+      "loss": 0.68,
+      "step": 3997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.17442448073718e-05,
+      "loss": 0.5653,
+      "step": 3998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1740567717130913e-05,
+      "loss": 1.0211,
+      "step": 3999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1736890384179347e-05,
+      "loss": 0.7491,
+      "step": 4000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.173321280902988e-05,
+      "loss": 0.9461,
+      "step": 4001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1729534992195326e-05,
+      "loss": 0.7979,
+      "step": 4002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1725856934188531e-05,
+      "loss": 0.594,
+      "step": 4003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.172217863552238e-05,
+      "loss": 0.7221,
+      "step": 4004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1718500096709783e-05,
+      "loss": 1.1065,
+      "step": 4005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1714821318263696e-05,
+      "loss": 0.7701,
+      "step": 4006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.171114230069709e-05,
+      "loss": 0.6902,
+      "step": 4007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1707463044522984e-05,
+      "loss": 0.7166,
+      "step": 4008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1703783550254429e-05,
+      "loss": 0.8175,
+      "step": 4009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1700103818404502e-05,
+      "loss": 1.0708,
+      "step": 4010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1696423849486319e-05,
+      "loss": 0.8369,
+      "step": 4011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1692743644013023e-05,
+      "loss": 0.614,
+      "step": 4012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1689063202497802e-05,
+      "loss": 0.7635,
+      "step": 4013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1685382525453867e-05,
+      "loss": 0.6157,
+      "step": 4014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.168170161339446e-05,
+      "loss": 0.6298,
+      "step": 4015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1678020466832866e-05,
+      "loss": 0.7428,
+      "step": 4016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.167433908628239e-05,
+      "loss": 0.8859,
+      "step": 4017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1670657472256383e-05,
+      "loss": 0.8104,
+      "step": 4018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1666975625268216e-05,
+      "loss": 0.9053,
+      "step": 4019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1663293545831302e-05,
+      "loss": 0.6562,
+      "step": 4020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.165961123445908e-05,
+      "loss": 0.6992,
+      "step": 4021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1655928691665026e-05,
+      "loss": 0.6061,
+      "step": 4022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1652245917962646e-05,
+      "loss": 0.96,
+      "step": 4023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1648562913865478e-05,
+      "loss": 0.7246,
+      "step": 4024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.164487967988709e-05,
+      "loss": 0.5697,
+      "step": 4025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.164119621654109e-05,
+      "loss": 0.7555,
+      "step": 4026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.163751252434111e-05,
+      "loss": 0.7258,
+      "step": 4027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1633828603800816e-05,
+      "loss": 0.7671,
+      "step": 4028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.163014445543391e-05,
+      "loss": 0.8937,
+      "step": 4029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1626460079754114e-05,
+      "loss": 0.8135,
+      "step": 4030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1622775477275197e-05,
+      "loss": 0.6953,
+      "step": 4031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1619090648510948e-05,
+      "loss": 0.6352,
+      "step": 4032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.16154055939752e-05,
+      "loss": 0.8199,
+      "step": 4033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.16117203141818e-05,
+      "loss": 0.6947,
+      "step": 4034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1608034809644645e-05,
+      "loss": 0.7082,
+      "step": 4035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1604349080877644e-05,
+      "loss": 1.0518,
+      "step": 4036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1600663128394756e-05,
+      "loss": 0.6673,
+      "step": 4037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1596976952709957e-05,
+      "loss": 0.6866,
+      "step": 4038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1593290554337267e-05,
+      "loss": 0.6633,
+      "step": 4039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.158960393379072e-05,
+      "loss": 0.6606,
+      "step": 4040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1585917091584397e-05,
+      "loss": 0.7439,
+      "step": 4041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1582230028232407e-05,
+      "loss": 0.6214,
+      "step": 4042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.157854274424888e-05,
+      "loss": 0.609,
+      "step": 4043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1574855240147986e-05,
+      "loss": 0.6149,
+      "step": 4044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1571167516443925e-05,
+      "loss": 0.7141,
+      "step": 4045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1567479573650924e-05,
+      "loss": 0.5748,
+      "step": 4046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1563791412283241e-05,
+      "loss": 0.7855,
+      "step": 4047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1560103032855167e-05,
+      "loss": 1.0288,
+      "step": 4048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.155641443588102e-05,
+      "loss": 0.9436,
+      "step": 4049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1552725621875156e-05,
+      "loss": 0.6341,
+      "step": 4050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1549036591351952e-05,
+      "loss": 0.779,
+      "step": 4051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.154534734482582e-05,
+      "loss": 0.9384,
+      "step": 4052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.15416578828112e-05,
+      "loss": 0.8185,
+      "step": 4053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1537968205822563e-05,
+      "loss": 0.6975,
+      "step": 4054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1534278314374412e-05,
+      "loss": 0.6842,
+      "step": 4055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1530588208981276e-05,
+      "loss": 0.6016,
+      "step": 4056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1526897890157715e-05,
+      "loss": 1.0091,
+      "step": 4057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1523207358418325e-05,
+      "loss": 0.7671,
+      "step": 4058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.151951661427772e-05,
+      "loss": 0.7892,
+      "step": 4059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1515825658250553e-05,
+      "loss": 1.0164,
+      "step": 4060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1512134490851503e-05,
+      "loss": 0.6718,
+      "step": 4061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1508443112595277e-05,
+      "loss": 0.685,
+      "step": 4062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1504751523996617e-05,
+      "loss": 0.8896,
+      "step": 4063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.150105972557029e-05,
+      "loss": 0.6641,
+      "step": 4064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1497367717831091e-05,
+      "loss": 0.6181,
+      "step": 4065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1493675501293846e-05,
+      "loss": 0.8952,
+      "step": 4066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1489983076473413e-05,
+      "loss": 0.6185,
+      "step": 4067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1486290443884666e-05,
+      "loss": 1.0621,
+      "step": 4068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1482597604042536e-05,
+      "loss": 0.7031,
+      "step": 4069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1478904557461951e-05,
+      "loss": 0.7665,
+      "step": 4070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1475211304657892e-05,
+      "loss": 0.7011,
+      "step": 4071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1471517846145348e-05,
+      "loss": 0.6839,
+      "step": 4072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1467824182439355e-05,
+      "loss": 0.6687,
+      "step": 4073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1464130314054966e-05,
+      "loss": 0.768,
+      "step": 4074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1460436241507274e-05,
+      "loss": 0.7205,
+      "step": 4075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1456741965311382e-05,
+      "loss": 0.6755,
+      "step": 4076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1453047485982438e-05,
+      "loss": 1.1072,
+      "step": 4077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1449352804035611e-05,
+      "loss": 0.5927,
+      "step": 4078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1445657919986108e-05,
+      "loss": 0.6984,
+      "step": 4079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1441962834349145e-05,
+      "loss": 0.6552,
+      "step": 4080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1438267547639985e-05,
+      "loss": 0.9367,
+      "step": 4081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1434572060373905e-05,
+      "loss": 0.6001,
+      "step": 4082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1430876373066224e-05,
+      "loss": 1.0374,
+      "step": 4083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1427180486232274e-05,
+      "loss": 0.8288,
+      "step": 4084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1423484400387427e-05,
+      "loss": 0.6715,
+      "step": 4085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1419788116047076e-05,
+      "loss": 0.6398,
+      "step": 4086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.141609163372664e-05,
+      "loss": 0.5963,
+      "step": 4087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1412394953941575e-05,
+      "loss": 0.6736,
+      "step": 4088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1408698077207354e-05,
+      "loss": 0.6202,
+      "step": 4089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1405001004039486e-05,
+      "loss": 0.7563,
+      "step": 4090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1401303734953502e-05,
+      "loss": 0.6941,
+      "step": 4091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1397606270464961e-05,
+      "loss": 1.0679,
+      "step": 4092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1393908611089448e-05,
+      "loss": 0.6957,
+      "step": 4093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.139021075734258e-05,
+      "loss": 0.9649,
+      "step": 4094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1386512709739996e-05,
+      "loss": 0.767,
+      "step": 4095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.138281446879737e-05,
+      "loss": 0.6622,
+      "step": 4096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.137911603503039e-05,
+      "loss": 0.7436,
+      "step": 4097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1375417408954786e-05,
+      "loss": 0.6489,
+      "step": 4098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1371718591086298e-05,
+      "loss": 0.7401,
+      "step": 4099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.136801958194071e-05,
+      "loss": 0.97,
+      "step": 4100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1364320382033818e-05,
+      "loss": 0.7925,
+      "step": 4101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360620991881455e-05,
+      "loss": 0.7292,
+      "step": 4102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1356921411999474e-05,
+      "loss": 0.5446,
+      "step": 4103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1353221642903756e-05,
+      "loss": 0.9165,
+      "step": 4104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1349521685110217e-05,
+      "loss": 0.6669,
+      "step": 4105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1345821539134782e-05,
+      "loss": 0.9198,
+      "step": 4106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1342121205493419e-05,
+      "loss": 0.8568,
+      "step": 4107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.133842068470211e-05,
+      "loss": 0.6525,
+      "step": 4108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1334719977276874e-05,
+      "loss": 0.6025,
+      "step": 4109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1331019083733743e-05,
+      "loss": 0.5953,
+      "step": 4110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1327318004588787e-05,
+      "loss": 0.9071,
+      "step": 4111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1323616740358097e-05,
+      "loss": 0.6697,
+      "step": 4112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1319915291557788e-05,
+      "loss": 0.6654,
+      "step": 4113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1316213658704e-05,
+      "loss": 0.6545,
+      "step": 4114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.131251184231291e-05,
+      "loss": 0.9742,
+      "step": 4115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1308809842900703e-05,
+      "loss": 0.7909,
+      "step": 4116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1305107660983606e-05,
+      "loss": 0.5955,
+      "step": 4117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1301405297077857e-05,
+      "loss": 0.6984,
+      "step": 4118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.129770275169973e-05,
+      "loss": 0.6839,
+      "step": 4119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1294000025365519e-05,
+      "loss": 0.7337,
+      "step": 4120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1290297118591546e-05,
+      "loss": 0.6683,
+      "step": 4121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1286594031894153e-05,
+      "loss": 0.6226,
+      "step": 4122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1282890765789719e-05,
+      "loss": 0.9452,
+      "step": 4123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.127918732079463e-05,
+      "loss": 1.0923,
+      "step": 4124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1275483697425315e-05,
+      "loss": 0.6905,
+      "step": 4125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1271779896198218e-05,
+      "loss": 0.6517,
+      "step": 4126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1268075917629807e-05,
+      "loss": 0.888,
+      "step": 4127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.126437176223658e-05,
+      "loss": 0.6164,
+      "step": 4128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1260667430535057e-05,
+      "loss": 0.7126,
+      "step": 4129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.125696292304178e-05,
+      "loss": 0.9301,
+      "step": 4130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.125325824027332e-05,
+      "loss": 0.8533,
+      "step": 4131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.124955338274627e-05,
+      "loss": 0.755,
+      "step": 4132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1245848350977246e-05,
+      "loss": 1.0098,
+      "step": 4133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1242143145482897e-05,
+      "loss": 0.6741,
+      "step": 4134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1238437766779881e-05,
+      "loss": 0.8289,
+      "step": 4135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1234732215384897e-05,
+      "loss": 1.0221,
+      "step": 4136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1231026491814652e-05,
+      "loss": 0.9273,
+      "step": 4137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1227320596585889e-05,
+      "loss": 0.774,
+      "step": 4138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1223614530215368e-05,
+      "loss": 0.6643,
+      "step": 4139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1219908293219878e-05,
+      "loss": 0.6221,
+      "step": 4140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1216201886116225e-05,
+      "loss": 0.6872,
+      "step": 4141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1212495309421252e-05,
+      "loss": 0.9051,
+      "step": 4142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1208788563651807e-05,
+      "loss": 1.0407,
+      "step": 4143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1205081649324778e-05,
+      "loss": 0.6258,
+      "step": 4144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1201374566957065e-05,
+      "loss": 0.713,
+      "step": 4145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1197667317065599e-05,
+      "loss": 0.5896,
+      "step": 4146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1193959900167329e-05,
+      "loss": 0.6326,
+      "step": 4147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1190252316779234e-05,
+      "loss": 0.6832,
+      "step": 4148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1186544567418307e-05,
+      "loss": 0.7645,
+      "step": 4149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1182836652601572e-05,
+      "loss": 0.6768,
+      "step": 4150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1179128572846074e-05,
+      "loss": 0.611,
+      "step": 4151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1175420328668878e-05,
+      "loss": 0.6693,
+      "step": 4152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1171711920587078e-05,
+      "loss": 0.7418,
+      "step": 4153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1168003349117779e-05,
+      "loss": 0.702,
+      "step": 4154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1164294614778129e-05,
+      "loss": 0.6359,
+      "step": 4155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1160585718085275e-05,
+      "loss": 0.7521,
+      "step": 4156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1156876659556407e-05,
+      "loss": 0.7693,
+      "step": 4157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1153167439708718e-05,
+      "loss": 0.7025,
+      "step": 4158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1149458059059446e-05,
+      "loss": 0.6398,
+      "step": 4159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.114574851812583e-05,
+      "loss": 0.9823,
+      "step": 4160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.114203881742515e-05,
+      "loss": 0.7797,
+      "step": 4161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1138328957474691e-05,
+      "loss": 0.792,
+      "step": 4162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1134618938791777e-05,
+      "loss": 0.877,
+      "step": 4163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1130908761893738e-05,
+      "loss": 0.6718,
+      "step": 4164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1127198427297936e-05,
+      "loss": 0.7183,
+      "step": 4165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1123487935521753e-05,
+      "loss": 0.808,
+      "step": 4166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1119777287082593e-05,
+      "loss": 0.6364,
+      "step": 4167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1116066482497882e-05,
+      "loss": 0.6286,
+      "step": 4168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1112355522285065e-05,
+      "loss": 0.6051,
+      "step": 4169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1108644406961616e-05,
+      "loss": 0.6168,
+      "step": 4170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1104933137045018e-05,
+      "loss": 0.6689,
+      "step": 4171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.110122171305279e-05,
+      "loss": 0.598,
+      "step": 4172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1097510135502458e-05,
+      "loss": 1.0884,
+      "step": 4173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1093798404911586e-05,
+      "loss": 0.7088,
+      "step": 4174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1090086521797743e-05,
+      "loss": 0.7789,
+      "step": 4175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1086374486678533e-05,
+      "loss": 0.6209,
+      "step": 4176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1082662300071566e-05,
+      "loss": 0.6979,
+      "step": 4177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1078949962494489e-05,
+      "loss": 0.6286,
+      "step": 4178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1075237474464961e-05,
+      "loss": 0.5966,
+      "step": 4179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1071524836500666e-05,
+      "loss": 1.0178,
+      "step": 4180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1067812049119304e-05,
+      "loss": 0.9884,
+      "step": 4181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1064099112838602e-05,
+      "loss": 0.6978,
+      "step": 4182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1060386028176297e-05,
+      "loss": 0.6193,
+      "step": 4183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1056672795650166e-05,
+      "loss": 0.9715,
+      "step": 4184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1052959415777983e-05,
+      "loss": 0.6301,
+      "step": 4185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.104924588907756e-05,
+      "loss": 0.6388,
+      "step": 4186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1045532216066721e-05,
+      "loss": 0.6059,
+      "step": 4187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.104181839726332e-05,
+      "loss": 0.7817,
+      "step": 4188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1038104433185215e-05,
+      "loss": 0.8993,
+      "step": 4189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1034390324350303e-05,
+      "loss": 0.5933,
+      "step": 4190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1030676071276484e-05,
+      "loss": 0.6507,
+      "step": 4191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1026961674481693e-05,
+      "loss": 0.7396,
+      "step": 4192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1023247134483875e-05,
+      "loss": 0.6287,
+      "step": 4193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1019532451800997e-05,
+      "loss": 0.8664,
+      "step": 4194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1015817626951043e-05,
+      "loss": 0.6347,
+      "step": 4195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1012102660452031e-05,
+      "loss": 0.6926,
+      "step": 4196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1008387552821981e-05,
+      "loss": 0.6563,
+      "step": 4197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1004672304578943e-05,
+      "loss": 0.7012,
+      "step": 4198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1000956916240985e-05,
+      "loss": 0.6928,
+      "step": 4199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0997241388326189e-05,
+      "loss": 0.8135,
+      "step": 4200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0993525721352668e-05,
+      "loss": 0.7014,
+      "step": 4201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0989809915838537e-05,
+      "loss": 1.1225,
+      "step": 4202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.098609397230195e-05,
+      "loss": 0.9023,
+      "step": 4203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0982377891261064e-05,
+      "loss": 0.6515,
+      "step": 4204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0978661673234067e-05,
+      "loss": 0.6445,
+      "step": 4205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0974945318739159e-05,
+      "loss": 0.6143,
+      "step": 4206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.097122882829456e-05,
+      "loss": 0.9236,
+      "step": 4207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0967512202418514e-05,
+      "loss": 0.7522,
+      "step": 4208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0963795441629275e-05,
+      "loss": 0.94,
+      "step": 4209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0960078546445124e-05,
+      "loss": 0.9907,
+      "step": 4210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0956361517384357e-05,
+      "loss": 0.6811,
+      "step": 4211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0952644354965287e-05,
+      "loss": 0.5738,
+      "step": 4212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0948927059706251e-05,
+      "loss": 0.6921,
+      "step": 4213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0945209632125599e-05,
+      "loss": 0.6096,
+      "step": 4214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0941492072741702e-05,
+      "loss": 0.6149,
+      "step": 4215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0937774382072952e-05,
+      "loss": 0.6419,
+      "step": 4216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.093405656063775e-05,
+      "loss": 0.8221,
+      "step": 4217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0930338608954529e-05,
+      "loss": 0.7329,
+      "step": 4218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.092662052754173e-05,
+      "loss": 0.597,
+      "step": 4219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0922902316917811e-05,
+      "loss": 0.658,
+      "step": 4220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0919183977601257e-05,
+      "loss": 0.824,
+      "step": 4221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0915465510110566e-05,
+      "loss": 0.6595,
+      "step": 4222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.091174691496425e-05,
+      "loss": 1.0117,
+      "step": 4223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0908028192680844e-05,
+      "loss": 0.653,
+      "step": 4224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.09043093437789e-05,
+      "loss": 0.5947,
+      "step": 4225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.090059036877699e-05,
+      "loss": 0.6688,
+      "step": 4226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0896871268193692e-05,
+      "loss": 0.9444,
+      "step": 4227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0893152042547618e-05,
+      "loss": 0.8633,
+      "step": 4228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0889432692357386e-05,
+      "loss": 0.6402,
+      "step": 4229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0885713218141638e-05,
+      "loss": 0.6973,
+      "step": 4230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0881993620419028e-05,
+      "loss": 0.5836,
+      "step": 4231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0878273899708227e-05,
+      "loss": 0.6286,
+      "step": 4232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0874554056527926e-05,
+      "loss": 0.6175,
+      "step": 4233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0870834091396836e-05,
+      "loss": 0.686,
+      "step": 4234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0867114004833681e-05,
+      "loss": 0.6547,
+      "step": 4235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0863393797357198e-05,
+      "loss": 0.6264,
+      "step": 4236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0859673469486151e-05,
+      "loss": 0.9839,
+      "step": 4237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0855953021739311e-05,
+      "loss": 0.6641,
+      "step": 4238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0852232454635475e-05,
+      "loss": 0.6608,
+      "step": 4239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0848511768693446e-05,
+      "loss": 0.6598,
+      "step": 4240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0844790964432052e-05,
+      "loss": 0.698,
+      "step": 4241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.084107004237013e-05,
+      "loss": 0.5839,
+      "step": 4242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0837349003026546e-05,
+      "loss": 1.0421,
+      "step": 4243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.083362784692017e-05,
+      "loss": 0.7227,
+      "step": 4244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0829906574569893e-05,
+      "loss": 0.6125,
+      "step": 4245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.082618518649462e-05,
+      "loss": 0.6014,
+      "step": 4246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0822463683213278e-05,
+      "loss": 0.5922,
+      "step": 4247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0818742065244803e-05,
+      "loss": 0.8053,
+      "step": 4248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0815020333108153e-05,
+      "loss": 0.582,
+      "step": 4249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0811298487322292e-05,
+      "loss": 0.6338,
+      "step": 4250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0807576528406216e-05,
+      "loss": 0.5624,
+      "step": 4251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0803854456878919e-05,
+      "loss": 0.624,
+      "step": 4252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.080013227325943e-05,
+      "loss": 0.7235,
+      "step": 4253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.079640997806677e-05,
+      "loss": 0.7413,
+      "step": 4254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0792687571820002e-05,
+      "loss": 0.8779,
+      "step": 4255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0788965055038179e-05,
+      "loss": 0.5995,
+      "step": 4256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.078524242824039e-05,
+      "loss": 0.7212,
+      "step": 4257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0781519691945723e-05,
+      "loss": 0.6352,
+      "step": 4258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0777796846673296e-05,
+      "loss": 0.6441,
+      "step": 4259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0774073892942231e-05,
+      "loss": 0.6409,
+      "step": 4260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0770350831271673e-05,
+      "loss": 0.7129,
+      "step": 4261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0766627662180775e-05,
+      "loss": 0.6266,
+      "step": 4262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0762904386188708e-05,
+      "loss": 0.8725,
+      "step": 4263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0759181003814662e-05,
+      "loss": 0.6412,
+      "step": 4264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0755457515577835e-05,
+      "loss": 0.6533,
+      "step": 4265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0751733921997444e-05,
+      "loss": 0.7084,
+      "step": 4266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0748010223592717e-05,
+      "loss": 0.8885,
+      "step": 4267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0744286420882906e-05,
+      "loss": 0.82,
+      "step": 4268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0740562514387262e-05,
+      "loss": 0.7364,
+      "step": 4269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0736838504625065e-05,
+      "loss": 0.6138,
+      "step": 4270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.07331143921156e-05,
+      "loss": 0.6642,
+      "step": 4271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0729390177378174e-05,
+      "loss": 0.7252,
+      "step": 4272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0725665860932101e-05,
+      "loss": 0.5916,
+      "step": 4273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0721941443296714e-05,
+      "loss": 1.024,
+      "step": 4274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0718216924991358e-05,
+      "loss": 0.7731,
+      "step": 4275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0714492306535393e-05,
+      "loss": 0.7292,
+      "step": 4276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.071076758844819e-05,
+      "loss": 0.5917,
+      "step": 4277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0707042771249139e-05,
+      "loss": 0.9477,
+      "step": 4278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.070331785545764e-05,
+      "loss": 0.5708,
+      "step": 4279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0699592841593105e-05,
+      "loss": 0.7372,
+      "step": 4280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.069586773017497e-05,
+      "loss": 0.5974,
+      "step": 4281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0692142521722673e-05,
+      "loss": 0.6893,
+      "step": 4282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.068841721675567e-05,
+      "loss": 0.6582,
+      "step": 4283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0684691815793427e-05,
+      "loss": 0.7749,
+      "step": 4284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.068096631935543e-05,
+      "loss": 0.5845,
+      "step": 4285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0677240727961175e-05,
+      "loss": 0.6337,
+      "step": 4286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0673515042130172e-05,
+      "loss": 1.0586,
+      "step": 4287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0669789262381939e-05,
+      "loss": 0.5558,
+      "step": 4288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0666063389236015e-05,
+      "loss": 0.6052,
+      "step": 4289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0662337423211947e-05,
+      "loss": 0.7825,
+      "step": 4290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06586113648293e-05,
+      "loss": 0.6082,
+      "step": 4291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0654885214607641e-05,
+      "loss": 0.638,
+      "step": 4292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0651158973066567e-05,
+      "loss": 0.7841,
+      "step": 4293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0647432640725667e-05,
+      "loss": 1.1319,
+      "step": 4294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0643706218104561e-05,
+      "loss": 0.7572,
+      "step": 4295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.063997970572287e-05,
+      "loss": 0.6494,
+      "step": 4296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0636253104100236e-05,
+      "loss": 0.5849,
+      "step": 4297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.06325264137563e-05,
+      "loss": 0.7126,
+      "step": 4298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0628799635210735e-05,
+      "loss": 0.5708,
+      "step": 4299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0625072768983211e-05,
+      "loss": 1.3094,
+      "step": 4300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0621345815593416e-05,
+      "loss": 0.7188,
+      "step": 4301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0617618775561045e-05,
+      "loss": 0.8018,
+      "step": 4302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0613891649405816e-05,
+      "loss": 0.989,
+      "step": 4303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0610164437647445e-05,
+      "loss": 0.9076,
+      "step": 4304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0606437140805672e-05,
+      "loss": 0.6849,
+      "step": 4305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0602709759400242e-05,
+      "loss": 0.6126,
+      "step": 4306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0598982293950913e-05,
+      "loss": 0.6557,
+      "step": 4307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0595254744977458e-05,
+      "loss": 0.7399,
+      "step": 4308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0591527112999658e-05,
+      "loss": 0.5923,
+      "step": 4309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0587799398537305e-05,
+      "loss": 0.7916,
+      "step": 4310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.058407160211021e-05,
+      "loss": 0.5522,
+      "step": 4311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.058034372423818e-05,
+      "loss": 0.5782,
+      "step": 4312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.057661576544105e-05,
+      "loss": 0.7596,
+      "step": 4313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0572887726238657e-05,
+      "loss": 0.6145,
+      "step": 4314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0569159607150852e-05,
+      "loss": 0.7244,
+      "step": 4315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0565431408697495e-05,
+      "loss": 0.7338,
+      "step": 4316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.056170313139846e-05,
+      "loss": 1.0613,
+      "step": 4317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0557974775773634e-05,
+      "loss": 0.8309,
+      "step": 4318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0554246342342905e-05,
+      "loss": 1.0452,
+      "step": 4319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0550517831626184e-05,
+      "loss": 0.7146,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0546789244143383e-05,
+      "loss": 0.6818,
+      "step": 4321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0543060580414433e-05,
+      "loss": 0.699,
+      "step": 4322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0539331840959268e-05,
+      "loss": 0.6332,
+      "step": 4323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.053560302629784e-05,
+      "loss": 0.7364,
+      "step": 4324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0531874136950102e-05,
+      "loss": 0.6492,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0528145173436027e-05,
+      "loss": 0.8074,
+      "step": 4326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0524416136275595e-05,
+      "loss": 0.7702,
+      "step": 4327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0520687025988792e-05,
+      "loss": 0.6279,
+      "step": 4328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0516957843095621e-05,
+      "loss": 0.7395,
+      "step": 4329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.051322858811609e-05,
+      "loss": 1.0276,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0509499261570222e-05,
+      "loss": 0.7324,
+      "step": 4331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0505769863978042e-05,
+      "loss": 0.6815,
+      "step": 4332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0502040395859594e-05,
+      "loss": 0.9774,
+      "step": 4333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0498310857734925e-05,
+      "loss": 0.8437,
+      "step": 4334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0494581250124099e-05,
+      "loss": 0.6915,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0490851573547178e-05,
+      "loss": 0.8617,
+      "step": 4336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.048712182852425e-05,
+      "loss": 1.1175,
+      "step": 4337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0483392015575393e-05,
+      "loss": 0.8482,
+      "step": 4338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0479662135220714e-05,
+      "loss": 0.5688,
+      "step": 4339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0475932187980313e-05,
+      "loss": 0.692,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0472202174374312e-05,
+      "loss": 0.6594,
+      "step": 4341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0468472094922832e-05,
+      "loss": 0.8412,
+      "step": 4342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0464741950146016e-05,
+      "loss": 0.8513,
+      "step": 4343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0461011740563996e-05,
+      "loss": 0.9884,
+      "step": 4344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0457281466696933e-05,
+      "loss": 0.8531,
+      "step": 4345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0453551129064992e-05,
+      "loss": 0.6883,
+      "step": 4346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0449820728188334e-05,
+      "loss": 0.6245,
+      "step": 4347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0446090264587147e-05,
+      "loss": 0.8647,
+      "step": 4348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0442359738781618e-05,
+      "loss": 0.7918,
+      "step": 4349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0438629151291944e-05,
+      "loss": 0.8593,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0434898502638326e-05,
+      "loss": 0.5704,
+      "step": 4351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0431167793340986e-05,
+      "loss": 0.6282,
+      "step": 4352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0427437023920137e-05,
+      "loss": 0.6843,
+      "step": 4353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0423706194896021e-05,
+      "loss": 0.5577,
+      "step": 4354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0419975306788872e-05,
+      "loss": 0.6669,
+      "step": 4355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0416244360118942e-05,
+      "loss": 0.5219,
+      "step": 4356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0412513355406478e-05,
+      "loss": 0.5292,
+      "step": 4357
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0408782293171752e-05,
+      "loss": 0.6976,
+      "step": 4358
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0405051173935032e-05,
+      "loss": 0.9569,
+      "step": 4359
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0401319998216602e-05,
+      "loss": 0.548,
+      "step": 4360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0397588766536744e-05,
+      "loss": 0.7142,
+      "step": 4361
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0393857479415757e-05,
+      "loss": 0.8113,
+      "step": 4362
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0390126137373941e-05,
+      "loss": 0.4972,
+      "step": 4363
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0386394740931616e-05,
+      "loss": 0.6225,
+      "step": 4364
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0382663290609091e-05,
+      "loss": 0.4986,
+      "step": 4365
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.03789317869267e-05,
+      "loss": 0.613,
+      "step": 4366
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0375200230404768e-05,
+      "loss": 0.4995,
+      "step": 4367
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0371468621563642e-05,
+      "loss": 0.8127,
+      "step": 4368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0367736960923668e-05,
+      "loss": 0.5855,
+      "step": 4369
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0364005249005204e-05,
+      "loss": 0.5515,
+      "step": 4370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.036027348632861e-05,
+      "loss": 0.6359,
+      "step": 4371
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0356541673414253e-05,
+      "loss": 0.6521,
+      "step": 4372
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0352809810782519e-05,
+      "loss": 0.9165,
+      "step": 4373
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.034907789895378e-05,
+      "loss": 0.6713,
+      "step": 4374
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0345345938448435e-05,
+      "loss": 0.5878,
+      "step": 4375
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0341613929786875e-05,
+      "loss": 0.5601,
+      "step": 4376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0337881873489511e-05,
+      "loss": 0.5037,
+      "step": 4377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0334149770076747e-05,
+      "loss": 0.6858,
+      "step": 4378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0330417620069007e-05,
+      "loss": 0.7005,
+      "step": 4379
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0326685423986707e-05,
+      "loss": 0.6683,
+      "step": 4380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0322953182350279e-05,
+      "loss": 0.6935,
+      "step": 4381
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0319220895680164e-05,
+      "loss": 0.5431,
+      "step": 4382
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0315488564496804e-05,
+      "loss": 0.5758,
+      "step": 4383
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0311756189320644e-05,
+      "loss": 0.6478,
+      "step": 4384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0308023770672143e-05,
+      "loss": 0.8236,
+      "step": 4385
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0304291309071756e-05,
+      "loss": 0.7885,
+      "step": 4386
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0300558805039957e-05,
+      "loss": 0.8725,
+      "step": 4387
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0296826259097216e-05,
+      "loss": 0.5679,
+      "step": 4388
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0293093671764015e-05,
+      "loss": 0.6533,
+      "step": 4389
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0289361043560833e-05,
+      "loss": 0.5145,
+      "step": 4390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0285628375008162e-05,
+      "loss": 0.5538,
+      "step": 4391
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0281895666626502e-05,
+      "loss": 0.5644,
+      "step": 4392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0278162918936351e-05,
+      "loss": 0.6142,
+      "step": 4393
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.027443013245822e-05,
+      "loss": 0.5843,
+      "step": 4394
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0270697307712614e-05,
+      "loss": 0.6071,
+      "step": 4395
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0266964445220058e-05,
+      "loss": 0.6453,
+      "step": 4396
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0263231545501068e-05,
+      "loss": 0.8177,
+      "step": 4397
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0259498609076181e-05,
+      "loss": 0.635,
+      "step": 4398
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0255765636465922e-05,
+      "loss": 0.6986,
+      "step": 4399
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0252032628190835e-05,
+      "loss": 0.8414,
+      "step": 4400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0248299584771462e-05,
+      "loss": 0.6523,
+      "step": 4401
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.024456650672835e-05,
+      "loss": 0.4899,
+      "step": 4402
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0240833394582052e-05,
+      "loss": 0.6709,
+      "step": 4403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.023710024885313e-05,
+      "loss": 0.7212,
+      "step": 4404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0233367070062141e-05,
+      "loss": 0.5982,
+      "step": 4405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0229633858729655e-05,
+      "loss": 0.6285,
+      "step": 4406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0225900615376243e-05,
+      "loss": 0.5597,
+      "step": 4407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.022216734052248e-05,
+      "loss": 0.5844,
+      "step": 4408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0218434034688947e-05,
+      "loss": 0.5817,
+      "step": 4409
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0214700698396231e-05,
+      "loss": 0.6487,
+      "step": 4410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0210967332164921e-05,
+      "loss": 0.5984,
+      "step": 4411
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0207233936515608e-05,
+      "loss": 0.6242,
+      "step": 4412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0203500511968888e-05,
+      "loss": 0.8531,
+      "step": 4413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0199767059045369e-05,
+      "loss": 0.5866,
+      "step": 4414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.019603357826565e-05,
+      "loss": 0.7379,
+      "step": 4415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0192300070150343e-05,
+      "loss": 0.6366,
+      "step": 4416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.018856653522006e-05,
+      "loss": 0.8221,
+      "step": 4417
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0184832973995416e-05,
+      "loss": 0.8316,
+      "step": 4418
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0181099386997038e-05,
+      "loss": 0.6348,
+      "step": 4419
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0177365774745542e-05,
+      "loss": 0.5851,
+      "step": 4420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0173632137761562e-05,
+      "loss": 0.6295,
+      "step": 4421
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0169898476565725e-05,
+      "loss": 0.5279,
+      "step": 4422
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0166164791678667e-05,
+      "loss": 0.6235,
+      "step": 4423
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0162431083621025e-05,
+      "loss": 0.5433,
+      "step": 4424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0158697352913443e-05,
+      "loss": 0.8777,
+      "step": 4425
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.015496360007656e-05,
+      "loss": 0.6255,
+      "step": 4426
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0151229825631026e-05,
+      "loss": 0.6885,
+      "step": 4427
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.014749603009749e-05,
+      "loss": 0.6114,
+      "step": 4428
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0143762213996612e-05,
+      "loss": 0.6087,
+      "step": 4429
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0140028377849034e-05,
+      "loss": 0.896,
+      "step": 4430
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0136294522175429e-05,
+      "loss": 0.9086,
+      "step": 4431
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.013256064749645e-05,
+      "loss": 0.6799,
+      "step": 4432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0128826754332763e-05,
+      "loss": 0.8986,
+      "step": 4433
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0125092843205038e-05,
+      "loss": 0.5164,
+      "step": 4434
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0121358914633942e-05,
+      "loss": 0.9185,
+      "step": 4435
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0117624969140141e-05,
+      "loss": 0.6219,
+      "step": 4436
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0113891007244316e-05,
+      "loss": 0.6304,
+      "step": 4437
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0110157029467147e-05,
+      "loss": 0.6186,
+      "step": 4438
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0106423036329302e-05,
+      "loss": 0.7046,
+      "step": 4439
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.010268902835147e-05,
+      "loss": 0.6602,
+      "step": 4440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.009895500605433e-05,
+      "loss": 0.7638,
+      "step": 4441
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0095220969958567e-05,
+      "loss": 0.6315,
+      "step": 4442
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0091486920584867e-05,
+      "loss": 0.641,
+      "step": 4443
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0087752858453923e-05,
+      "loss": 0.7091,
+      "step": 4444
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0084018784086418e-05,
+      "loss": 0.7343,
+      "step": 4445
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0080284698003052e-05,
+      "loss": 0.5307,
+      "step": 4446
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.007655060072451e-05,
+      "loss": 0.5883,
+      "step": 4447
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0072816492771498e-05,
+      "loss": 0.6052,
+      "step": 4448
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0069082374664703e-05,
+      "loss": 0.5493,
+      "step": 4449
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0065348246924829e-05,
+      "loss": 0.4624,
+      "step": 4450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.006161411007257e-05,
+      "loss": 0.8436,
+      "step": 4451
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0057879964628634e-05,
+      "loss": 0.6545,
+      "step": 4452
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0054145811113716e-05,
+      "loss": 0.8666,
+      "step": 4453
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0050411650048522e-05,
+      "loss": 0.6021,
+      "step": 4454
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.004667748195376e-05,
+      "loss": 0.5688,
+      "step": 4455
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0042943307350126e-05,
+      "loss": 0.6966,
+      "step": 4456
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0039209126758333e-05,
+      "loss": 0.4916,
+      "step": 4457
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0035474940699089e-05,
+      "loss": 0.6872,
+      "step": 4458
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0031740749693095e-05,
+      "loss": 0.5543,
+      "step": 4459
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0028006554261065e-05,
+      "loss": 0.5616,
+      "step": 4460
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0024272354923705e-05,
+      "loss": 0.5044,
+      "step": 4461
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0020538152201725e-05,
+      "loss": 0.5452,
+      "step": 4462
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0016803946615835e-05,
+      "loss": 0.6006,
+      "step": 4463
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0013069738686745e-05,
+      "loss": 0.5616,
+      "step": 4464
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0009335528935169e-05,
+      "loss": 0.673,
+      "step": 4465
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0005601317881813e-05,
+      "loss": 0.7144,
+      "step": 4466
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0001867106047391e-05,
+      "loss": 0.5182,
+      "step": 4467
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99813289395261e-06,
+      "loss": 0.6446,
+      "step": 4468
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.994398682118189e-06,
+      "loss": 0.7088,
+      "step": 4469
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.990664471064833e-06,
+      "loss": 0.5777,
+      "step": 4470
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.986930261313255e-06,
+      "loss": 0.6655,
+      "step": 4471
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.983196053384167e-06,
+      "loss": 0.5379,
+      "step": 4472
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.979461847798279e-06,
+      "loss": 0.8618,
+      "step": 4473
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.975727645076297e-06,
+      "loss": 0.6358,
+      "step": 4474
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.971993445738938e-06,
+      "loss": 0.5445,
+      "step": 4475
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.968259250306908e-06,
+      "loss": 0.6516,
+      "step": 4476
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.964525059300917e-06,
+      "loss": 0.5534,
+      "step": 4477
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.960790873241668e-06,
+      "loss": 0.5478,
+      "step": 4478
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.957056692649878e-06,
+      "loss": 0.6383,
+      "step": 4479
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.953322518046245e-06,
+      "loss": 0.7675,
+      "step": 4480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.949588349951482e-06,
+      "loss": 0.6377,
+      "step": 4481
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.945854188886287e-06,
+      "loss": 0.5057,
+      "step": 4482
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.942120035371371e-06,
+      "loss": 0.8383,
+      "step": 4483
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.938385889927433e-06,
+      "loss": 0.9451,
+      "step": 4484
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.934651753075176e-06,
+      "loss": 0.7786,
+      "step": 4485
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.930917625335297e-06,
+      "loss": 0.5934,
+      "step": 4486
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.927183507228503e-06,
+      "loss": 0.7002,
+      "step": 4487
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.923449399275488e-06,
+      "loss": 0.9473,
+      "step": 4488
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.91971530199695e-06,
+      "loss": 0.5896,
+      "step": 4489
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.915981215913583e-06,
+      "loss": 0.5371,
+      "step": 4490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.91224714154608e-06,
+      "loss": 0.5714,
+      "step": 4491
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.908513079415135e-06,
+      "loss": 0.6557,
+      "step": 4492
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.904779030041434e-06,
+      "loss": 0.5791,
+      "step": 4493
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.901044993945675e-06,
+      "loss": 0.6983,
+      "step": 4494
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.897310971648533e-06,
+      "loss": 0.5578,
+      "step": 4495
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.893576963670701e-06,
+      "loss": 0.6266,
+      "step": 4496
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.889842970532856e-06,
+      "loss": 0.6561,
+      "step": 4497
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.886108992755686e-06,
+      "loss": 0.8234,
+      "step": 4498
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.88237503085986e-06,
+      "loss": 0.5891,
+      "step": 4499
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.878641085366065e-06,
+      "loss": 0.6729,
+      "step": 4500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.874907156794967e-06,
+      "loss": 0.6411,
+      "step": 4501
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.87117324566724e-06,
+      "loss": 0.8895,
+      "step": 4502
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.867439352503554e-06,
+      "loss": 0.632,
+      "step": 4503
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.863705477824576e-06,
+      "loss": 0.7226,
+      "step": 4504
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.859971622150966e-06,
+      "loss": 0.6218,
+      "step": 4505
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.85623778600339e-06,
+      "loss": 0.477,
+      "step": 4506
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.85250396990251e-06,
+      "loss": 0.7846,
+      "step": 4507
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.848770174368975e-06,
+      "loss": 0.6827,
+      "step": 4508
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.845036399923441e-06,
+      "loss": 0.8649,
+      "step": 4509
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.841302647086558e-06,
+      "loss": 0.5311,
+      "step": 4510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.837568916378977e-06,
+      "loss": 0.5681,
+      "step": 4511
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.833835208321336e-06,
+      "loss": 0.5684,
+      "step": 4512
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.83010152343428e-06,
+      "loss": 0.5882,
+      "step": 4513
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.82636786223844e-06,
+      "loss": 0.5516,
+      "step": 4514
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.822634225254461e-06,
+      "loss": 0.5524,
+      "step": 4515
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.818900613002967e-06,
+      "loss": 0.5424,
+      "step": 4516
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.815167026004588e-06,
+      "loss": 0.5571,
+      "step": 4517
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.811433464779945e-06,
+      "loss": 0.844,
+      "step": 4518
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.807699929849662e-06,
+      "loss": 0.5876,
+      "step": 4519
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.803966421734355e-06,
+      "loss": 0.5082,
+      "step": 4520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.800232940954636e-06,
+      "loss": 0.6282,
+      "step": 4521
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.796499488031113e-06,
+      "loss": 0.6222,
+      "step": 4522
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792766063484393e-06,
+      "loss": 0.5722,
+      "step": 4523
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.78903266783508e-06,
+      "loss": 0.529,
+      "step": 4524
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.785299301603769e-06,
+      "loss": 0.4942,
+      "step": 4525
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.781565965311053e-06,
+      "loss": 0.625,
+      "step": 4526
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.777832659477522e-06,
+      "loss": 0.5842,
+      "step": 4527
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.77409938462376e-06,
+      "loss": 0.5121,
+      "step": 4528
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.770366141270348e-06,
+      "loss": 0.8633,
+      "step": 4529
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.766632929937862e-06,
+      "loss": 0.8525,
+      "step": 4530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.762899751146873e-06,
+      "loss": 0.6604,
+      "step": 4531
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.75916660541795e-06,
+      "loss": 0.5258,
+      "step": 4532
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.755433493271653e-06,
+      "loss": 0.8571,
+      "step": 4533
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.751700415228541e-06,
+      "loss": 0.6457,
+      "step": 4534
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.747967371809168e-06,
+      "loss": 0.6402,
+      "step": 4535
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.744234363534081e-06,
+      "loss": 0.558,
+      "step": 4536
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.740501390923824e-06,
+      "loss": 0.5366,
+      "step": 4537
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.736768454498935e-06,
+      "loss": 0.5203,
+      "step": 4538
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.733035554779947e-06,
+      "loss": 0.7127,
+      "step": 4539
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.729302692287391e-06,
+      "loss": 0.5733,
+      "step": 4540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.725569867541786e-06,
+      "loss": 0.8118,
+      "step": 4541
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.721837081063649e-06,
+      "loss": 0.6417,
+      "step": 4542
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.7181043333735e-06,
+      "loss": 0.808,
+      "step": 4543
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.71437162499184e-06,
+      "loss": 0.5406,
+      "step": 4544
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.71063895643917e-06,
+      "loss": 0.6691,
+      "step": 4545
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.706906328235987e-06,
+      "loss": 0.6265,
+      "step": 4546
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.703173740902785e-06,
+      "loss": 0.5302,
+      "step": 4547
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.699441194960045e-06,
+      "loss": 0.8696,
+      "step": 4548
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.695708690928246e-06,
+      "loss": 0.6123,
+      "step": 4549
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.69197622932786e-06,
+      "loss": 0.5603,
+      "step": 4550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.68824381067936e-06,
+      "loss": 0.5288,
+      "step": 4551
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.684511435503198e-06,
+      "loss": 0.5655,
+      "step": 4552
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.680779104319839e-06,
+      "loss": 0.6667,
+      "step": 4553
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.677046817649723e-06,
+      "loss": 0.7373,
+      "step": 4554
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.6733145760133e-06,
+      "loss": 0.6398,
+      "step": 4555
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.669582379931e-06,
+      "loss": 0.8817,
+      "step": 4556
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.665850229923258e-06,
+      "loss": 0.531,
+      "step": 4557
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.662118126510494e-06,
+      "loss": 0.8344,
+      "step": 4558
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.65838607021313e-06,
+      "loss": 0.5297,
+      "step": 4559
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.654654061551569e-06,
+      "loss": 0.5253,
+      "step": 4560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.650922101046222e-06,
+      "loss": 0.6548,
+      "step": 4561
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.647190189217486e-06,
+      "loss": 0.8686,
+      "step": 4562
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.643458326585747e-06,
+      "loss": 0.7104,
+      "step": 4563
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.639726513671394e-06,
+      "loss": 0.6329,
+      "step": 4564
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.635994750994797e-06,
+      "loss": 0.5682,
+      "step": 4565
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.632263039076333e-06,
+      "loss": 0.6447,
+      "step": 4566
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.62853137843636e-06,
+      "loss": 0.6109,
+      "step": 4567
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.624799769595234e-06,
+      "loss": 0.5323,
+      "step": 4568
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.621068213073304e-06,
+      "loss": 0.5722,
+      "step": 4569
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.61733670939091e-06,
+      "loss": 0.6525,
+      "step": 4570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.613605259068387e-06,
+      "loss": 0.5743,
+      "step": 4571
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.60987386262606e-06,
+      "loss": 0.9653,
+      "step": 4572
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.606142520584248e-06,
+      "loss": 0.6908,
+      "step": 4573
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.602411233463263e-06,
+      "loss": 0.5832,
+      "step": 4574
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.598680001783405e-06,
+      "loss": 0.8022,
+      "step": 4575
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.594948826064973e-06,
+      "loss": 0.8261,
+      "step": 4576
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.591217706828253e-06,
+      "loss": 0.6929,
+      "step": 4577
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.587486644593524e-06,
+      "loss": 0.4576,
+      "step": 4578
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.58375563988106e-06,
+      "loss": 0.5186,
+      "step": 4579
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.58002469321113e-06,
+      "loss": 0.6799,
+      "step": 4580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.576293805103979e-06,
+      "loss": 0.6383,
+      "step": 4581
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.572562976079863e-06,
+      "loss": 0.6711,
+      "step": 4582
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.568832206659017e-06,
+      "loss": 0.5587,
+      "step": 4583
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.565101497361677e-06,
+      "loss": 0.6587,
+      "step": 4584
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.561370848708061e-06,
+      "loss": 0.5946,
+      "step": 4585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.557640261218385e-06,
+      "loss": 0.571,
+      "step": 4586
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.553909735412854e-06,
+      "loss": 0.8251,
+      "step": 4587
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.550179271811668e-06,
+      "loss": 0.8424,
+      "step": 4588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.546448870935013e-06,
+      "loss": 0.7914,
+      "step": 4589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.542718533303069e-06,
+      "loss": 0.5145,
+      "step": 4590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.538988259436005e-06,
+      "loss": 0.5337,
+      "step": 4591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.535258049853991e-06,
+      "loss": 0.7015,
+      "step": 4592
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.53152790507717e-06,
+      "loss": 0.9529,
+      "step": 4593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.527797825625693e-06,
+      "loss": 0.9579,
+      "step": 4594
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.524067812019689e-06,
+      "loss": 0.7043,
+      "step": 4595
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.520337864779291e-06,
+      "loss": 0.579,
+      "step": 4596
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.516607984424607e-06,
+      "loss": 0.5631,
+      "step": 4597
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.512878171475752e-06,
+      "loss": 0.5105,
+      "step": 4598
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.509148426452822e-06,
+      "loss": 0.5928,
+      "step": 4599
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.505418749875903e-06,
+      "loss": 0.6679,
+      "step": 4600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.501689142265076e-06,
+      "loss": 0.6588,
+      "step": 4601
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.497959604140408e-06,
+      "loss": 0.71,
+      "step": 4602
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.494230136021962e-06,
+      "loss": 0.997,
+      "step": 4603
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.490500738429783e-06,
+      "loss": 0.8608,
+      "step": 4604
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.486771411883914e-06,
+      "loss": 0.5828,
+      "step": 4605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.483042156904382e-06,
+      "loss": 0.619,
+      "step": 4606
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.479312974011213e-06,
+      "loss": 0.5328,
+      "step": 4607
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.475583863724409e-06,
+      "loss": 0.7802,
+      "step": 4608
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.471854826563978e-06,
+      "loss": 0.5988,
+      "step": 4609
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.468125863049902e-06,
+      "loss": 0.8076,
+      "step": 4610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.464396973702165e-06,
+      "loss": 0.8119,
+      "step": 4611
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.460668159040734e-06,
+      "loss": 0.6305,
+      "step": 4612
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.45693941958557e-06,
+      "loss": 0.8848,
+      "step": 4613
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.45321075585662e-06,
+      "loss": 0.6292,
+      "step": 4614
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.449482168373816e-06,
+      "loss": 0.5685,
+      "step": 4615
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.445753657657097e-06,
+      "loss": 0.4641,
+      "step": 4616
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.442025224226367e-06,
+      "loss": 0.5959,
+      "step": 4617
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.438296868601542e-06,
+      "loss": 0.5472,
+      "step": 4618
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.434568591302506e-06,
+      "loss": 0.6505,
+      "step": 4619
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.430840392849152e-06,
+      "loss": 0.5829,
+      "step": 4620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.427112273761344e-06,
+      "loss": 0.9349,
+      "step": 4621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.423384234558952e-06,
+      "loss": 0.5774,
+      "step": 4622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.419656275761822e-06,
+      "loss": 0.5761,
+      "step": 4623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.415928397889796e-06,
+      "loss": 0.8979,
+      "step": 4624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.412200601462696e-06,
+      "loss": 0.5089,
+      "step": 4625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.408472887000346e-06,
+      "loss": 0.5685,
+      "step": 4626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.404745255022543e-06,
+      "loss": 0.6992,
+      "step": 4627
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.401017706049089e-06,
+      "loss": 0.5579,
+      "step": 4628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.397290240599761e-06,
+      "loss": 0.8183,
+      "step": 4629
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.393562859194332e-06,
+      "loss": 0.6832,
+      "step": 4630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.389835562352558e-06,
+      "loss": 0.9105,
+      "step": 4631
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.38610835059419e-06,
+      "loss": 0.6366,
+      "step": 4632
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.382381224438959e-06,
+      "loss": 0.6437,
+      "step": 4633
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.378654184406585e-06,
+      "loss": 0.9845,
+      "step": 4634
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.37492723101679e-06,
+      "loss": 0.5907,
+      "step": 4635
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.371200364789265e-06,
+      "loss": 0.6889,
+      "step": 4636
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.367473586243701e-06,
+      "loss": 0.7804,
+      "step": 4637
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.363746895899768e-06,
+      "loss": 0.8207,
+      "step": 4638
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.360020294277132e-06,
+      "loss": 0.7558,
+      "step": 4639
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.35629378189544e-06,
+      "loss": 0.6553,
+      "step": 4640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.352567359274336e-06,
+      "loss": 0.9048,
+      "step": 4641
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.348841026933437e-06,
+      "loss": 0.7073,
+      "step": 4642
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.34511478539236e-06,
+      "loss": 0.5614,
+      "step": 4643
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.341388635170702e-06,
+      "loss": 0.8875,
+      "step": 4644
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.337662576788056e-06,
+      "loss": 0.6001,
+      "step": 4645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.333936610763987e-06,
+      "loss": 0.812,
+      "step": 4646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.330210737618066e-06,
+      "loss": 0.59,
+      "step": 4647
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.326484957869833e-06,
+      "loss": 0.5547,
+      "step": 4648
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32275927203883e-06,
+      "loss": 0.8193,
+      "step": 4649
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.319033680644575e-06,
+      "loss": 0.9611,
+      "step": 4650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.31530818420658e-06,
+      "loss": 0.4919,
+      "step": 4651
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.311582783244336e-06,
+      "loss": 0.7916,
+      "step": 4652
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.307857478277328e-06,
+      "loss": 0.6359,
+      "step": 4653
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.304132269825033e-06,
+      "loss": 0.537,
+      "step": 4654
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.300407158406895e-06,
+      "loss": 0.7107,
+      "step": 4655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.296682144542364e-06,
+      "loss": 0.4963,
+      "step": 4656
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.292957228750864e-06,
+      "loss": 0.8288,
+      "step": 4657
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.289232411551815e-06,
+      "loss": 0.5438,
+      "step": 4658
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.28550769346461e-06,
+      "loss": 0.9612,
+      "step": 4659
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.281783075008645e-06,
+      "loss": 0.6539,
+      "step": 4660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.278058556703287e-06,
+      "loss": 0.7851,
+      "step": 4661
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.274334139067902e-06,
+      "loss": 0.8854,
+      "step": 4662
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.270609822621827e-06,
+      "loss": 0.4451,
+      "step": 4663
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.266885607884402e-06,
+      "loss": 0.5922,
+      "step": 4664
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.263161495374939e-06,
+      "loss": 0.5873,
+      "step": 4665
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.259437485612743e-06,
+      "loss": 0.6448,
+      "step": 4666
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.2557135791171e-06,
+      "loss": 0.5454,
+      "step": 4667
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.251989776407286e-06,
+      "loss": 0.7239,
+      "step": 4668
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.24826607800256e-06,
+      "loss": 0.5616,
+      "step": 4669
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.24454248442217e-06,
+      "loss": 0.5746,
+      "step": 4670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.24081899618534e-06,
+      "loss": 0.627,
+      "step": 4671
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.237095613811292e-06,
+      "loss": 0.534,
+      "step": 4672
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.233372337819229e-06,
+      "loss": 0.7259,
+      "step": 4673
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.229649168728329e-06,
+      "loss": 0.6131,
+      "step": 4674
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.225926107057772e-06,
+      "loss": 0.5617,
+      "step": 4675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.222203153326705e-06,
+      "loss": 0.884,
+      "step": 4676
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.218480308054279e-06,
+      "loss": 0.5402,
+      "step": 4677
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.214757571759614e-06,
+      "loss": 0.724,
+      "step": 4678
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.211034944961825e-06,
+      "loss": 0.4983,
+      "step": 4679
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.207312428180002e-06,
+      "loss": 0.5377,
+      "step": 4680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.203590021933232e-06,
+      "loss": 0.7166,
+      "step": 4681
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.199867726740575e-06,
+      "loss": 0.5863,
+      "step": 4682
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.196145543121083e-06,
+      "loss": 0.5949,
+      "step": 4683
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.192423471593787e-06,
+      "loss": 0.5832,
+      "step": 4684
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.188701512677713e-06,
+      "loss": 0.6541,
+      "step": 4685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.184979666891854e-06,
+      "loss": 0.6714,
+      "step": 4686
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.181257934755202e-06,
+      "loss": 0.6432,
+      "step": 4687
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.177536316786726e-06,
+      "loss": 0.6165,
+      "step": 4688
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.173814813505381e-06,
+      "loss": 0.6055,
+      "step": 4689
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.170093425430107e-06,
+      "loss": 0.5879,
+      "step": 4690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.16637215307983e-06,
+      "loss": 0.6247,
+      "step": 4691
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.162650996973455e-06,
+      "loss": 0.5841,
+      "step": 4692
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.15892995762987e-06,
+      "loss": 0.8579,
+      "step": 4693
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.155209035567952e-06,
+      "loss": 0.7098,
+      "step": 4694
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.151488231306557e-06,
+      "loss": 0.6813,
+      "step": 4695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.147767545364527e-06,
+      "loss": 0.5763,
+      "step": 4696
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.14404697826069e-06,
+      "loss": 0.5888,
+      "step": 4697
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.14032653051385e-06,
+      "loss": 0.7125,
+      "step": 4698
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.136606202642805e-06,
+      "loss": 0.609,
+      "step": 4699
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.132885995166322e-06,
+      "loss": 0.6636,
+      "step": 4700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.129165908603167e-06,
+      "loss": 0.6021,
+      "step": 4701
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.125445943472077e-06,
+      "loss": 0.8908,
+      "step": 4702
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12172610029178e-06,
+      "loss": 0.6342,
+      "step": 4703
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.118006379580978e-06,
+      "loss": 0.5165,
+      "step": 4704
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.114286781858367e-06,
+      "loss": 0.534,
+      "step": 4705
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.110567307642616e-06,
+      "loss": 0.5064,
+      "step": 4706
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.10684795745238e-06,
+      "loss": 0.5512,
+      "step": 4707
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.103128731806308e-06,
+      "loss": 0.551,
+      "step": 4708
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.099409631223012e-06,
+      "loss": 0.5236,
+      "step": 4709
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.095690656221101e-06,
+      "loss": 0.5486,
+      "step": 4710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.091971807319156e-06,
+      "loss": 0.7097,
+      "step": 4711
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.088253085035752e-06,
+      "loss": 0.8596,
+      "step": 4712
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.084534489889435e-06,
+      "loss": 0.6739,
+      "step": 4713
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.080816022398744e-06,
+      "loss": 0.5804,
+      "step": 4714
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.07709768308219e-06,
+      "loss": 0.6089,
+      "step": 4715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.073379472458274e-06,
+      "loss": 0.773,
+      "step": 4716
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.069661391045473e-06,
+      "loss": 0.571,
+      "step": 4717
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.065943439362253e-06,
+      "loss": 0.4746,
+      "step": 4718
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.062225617927052e-06,
+      "loss": 0.6003,
+      "step": 4719
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.058507927258301e-06,
+      "loss": 0.5181,
+      "step": 4720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.054790367874404e-06,
+      "loss": 0.5637,
+      "step": 4721
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.051072940293754e-06,
+      "loss": 0.8162,
+      "step": 4722
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.047355645034715e-06,
+      "loss": 0.7394,
+      "step": 4723
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.043638482615648e-06,
+      "loss": 0.6686,
+      "step": 4724
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.03992145355488e-06,
+      "loss": 0.6092,
+      "step": 4725
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.036204558370725e-06,
+      "loss": 0.6482,
+      "step": 4726
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.032487797581488e-06,
+      "loss": 0.6333,
+      "step": 4727
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.02877117170544e-06,
+      "loss": 0.6006,
+      "step": 4728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.025054681260843e-06,
+      "loss": 0.6238,
+      "step": 4729
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.021338326765934e-06,
+      "loss": 0.7642,
+      "step": 4730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.017622108738939e-06,
+      "loss": 0.5917,
+      "step": 4731
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.013906027698054e-06,
+      "loss": 0.681,
+      "step": 4732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.010190084161466e-06,
+      "loss": 0.7514,
+      "step": 4733
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.006474278647337e-06,
+      "loss": 0.5742,
+      "step": 4734
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.002758611673812e-06,
+      "loss": 0.5658,
+      "step": 4735
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.999043083759016e-06,
+      "loss": 0.7074,
+      "step": 4736
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.995327695421059e-06,
+      "loss": 0.6092,
+      "step": 4737
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.99161244717802e-06,
+      "loss": 0.602,
+      "step": 4738
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.987897339547972e-06,
+      "loss": 0.8765,
+      "step": 4739
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.984182373048958e-06,
+      "loss": 0.5864,
+      "step": 4740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.980467548199008e-06,
+      "loss": 0.5839,
+      "step": 4741
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97675286551613e-06,
+      "loss": 0.4745,
+      "step": 4742
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97303832551831e-06,
+      "loss": 0.6815,
+      "step": 4743
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.969323928723515e-06,
+      "loss": 0.8154,
+      "step": 4744
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.965609675649699e-06,
+      "loss": 0.5788,
+      "step": 4745
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.961895566814785e-06,
+      "loss": 0.5514,
+      "step": 4746
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.958181602736682e-06,
+      "loss": 0.6087,
+      "step": 4747
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.954467783933279e-06,
+      "loss": 0.5927,
+      "step": 4748
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.950754110922442e-06,
+      "loss": 0.7001,
+      "step": 4749
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.94704058422202e-06,
+      "loss": 0.4112,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.94332720434984e-06,
+      "loss": 0.8729,
+      "step": 4751
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.939613971823705e-06,
+      "loss": 0.6827,
+      "step": 4752
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.935900887161401e-06,
+      "loss": 0.4965,
+      "step": 4753
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.932187950880698e-06,
+      "loss": 0.5866,
+      "step": 4754
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.928475163499337e-06,
+      "loss": 0.7161,
+      "step": 4755
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.924762525535042e-06,
+      "loss": 0.5708,
+      "step": 4756
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.921050037505513e-06,
+      "loss": 0.6508,
+      "step": 4757
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.917337699928439e-06,
+      "loss": 0.6246,
+      "step": 4758
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.913625513321474e-06,
+      "loss": 0.7541,
+      "step": 4759
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.909913478202262e-06,
+      "loss": 0.7284,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.906201595088419e-06,
+      "loss": 0.5171,
+      "step": 4761
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.902489864497547e-06,
+      "loss": 0.6495,
+      "step": 4762
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.898778286947214e-06,
+      "loss": 0.516,
+      "step": 4763
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.895066862954982e-06,
+      "loss": 0.617,
+      "step": 4764
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.891355593038388e-06,
+      "loss": 0.7011,
+      "step": 4765
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.887644477714935e-06,
+      "loss": 0.6004,
+      "step": 4766
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.88393351750212e-06,
+      "loss": 0.6342,
+      "step": 4767
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.880222712917408e-06,
+      "loss": 0.7891,
+      "step": 4768
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.87651206447825e-06,
+      "loss": 0.6305,
+      "step": 4769
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.872801572702066e-06,
+      "loss": 0.7118,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.869091238106267e-06,
+      "loss": 0.7671,
+      "step": 4771
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.865381061208227e-06,
+      "loss": 0.6102,
+      "step": 4772
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.861671042525312e-06,
+      "loss": 0.5737,
+      "step": 4773
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.857961182574854e-06,
+      "loss": 0.7335,
+      "step": 4774
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.854251481874172e-06,
+      "loss": 0.7187,
+      "step": 4775
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.85054194094056e-06,
+      "loss": 0.5381,
+      "step": 4776
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.846832560291287e-06,
+      "loss": 0.6451,
+      "step": 4777
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.8431233404436e-06,
+      "loss": 0.5341,
+      "step": 4778
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.83941428191473e-06,
+      "loss": 0.6702,
+      "step": 4779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.835705385221876e-06,
+      "loss": 0.6191,
+      "step": 4780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.83199665088222e-06,
+      "loss": 0.5146,
+      "step": 4781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.828288079412925e-06,
+      "loss": 0.5215,
+      "step": 4782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.824579671331122e-06,
+      "loss": 0.7146,
+      "step": 4783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.820871427153929e-06,
+      "loss": 0.6339,
+      "step": 4784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.81716334739843e-06,
+      "loss": 0.7292,
+      "step": 4785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.813455432581696e-06,
+      "loss": 0.5172,
+      "step": 4786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.80974768322077e-06,
+      "loss": 0.6585,
+      "step": 4787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.806040099832675e-06,
+      "loss": 0.5964,
+      "step": 4788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.802332682934405e-06,
+      "loss": 0.6437,
+      "step": 4789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.798625433042939e-06,
+      "loss": 0.7601,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.794918350675225e-06,
+      "loss": 0.5834,
+      "step": 4791
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.791211436348196e-06,
+      "loss": 0.8463,
+      "step": 4792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.787504690578751e-06,
+      "loss": 0.6393,
+      "step": 4793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.783798113883776e-06,
+      "loss": 0.8963,
+      "step": 4794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.780091706780127e-06,
+      "loss": 0.8496,
+      "step": 4795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.776385469784637e-06,
+      "loss": 0.6587,
+      "step": 4796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.772679403414118e-06,
+      "loss": 0.5823,
+      "step": 4797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.768973508185355e-06,
+      "loss": 0.5073,
+      "step": 4798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.765267784615107e-06,
+      "loss": 0.8573,
+      "step": 4799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.761562233220119e-06,
+      "loss": 0.5923,
+      "step": 4800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.757856854517106e-06,
+      "loss": 0.5917,
+      "step": 4801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.754151649022754e-06,
+      "loss": 0.8829,
+      "step": 4802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.750446617253733e-06,
+      "loss": 0.5595,
+      "step": 4803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.746741759726682e-06,
+      "loss": 0.5891,
+      "step": 4804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.743037076958224e-06,
+      "loss": 0.8344,
+      "step": 4805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.739332569464947e-06,
+      "loss": 0.6801,
+      "step": 4806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.735628237763422e-06,
+      "loss": 0.6632,
+      "step": 4807
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.731924082370194e-06,
+      "loss": 0.6861,
+      "step": 4808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.728220103801786e-06,
+      "loss": 0.6251,
+      "step": 4809
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.724516302574687e-06,
+      "loss": 0.6099,
+      "step": 4810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.720812679205372e-06,
+      "loss": 0.5268,
+      "step": 4811
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.717109234210286e-06,
+      "loss": 0.538,
+      "step": 4812
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.713405968105849e-06,
+      "loss": 0.8261,
+      "step": 4813
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.70970288140846e-06,
+      "loss": 0.6341,
+      "step": 4814
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.705999974634485e-06,
+      "loss": 0.5855,
+      "step": 4815
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.702297248300275e-06,
+      "loss": 0.5405,
+      "step": 4816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.698594702922147e-06,
+      "loss": 0.4991,
+      "step": 4817
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.694892339016396e-06,
+      "loss": 0.8467,
+      "step": 4818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.691190157099296e-06,
+      "loss": 0.6246,
+      "step": 4819
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.68748815768709e-06,
+      "loss": 0.6417,
+      "step": 4820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.683786341296e-06,
+      "loss": 0.4521,
+      "step": 4821
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.680084708442214e-06,
+      "loss": 0.7956,
+      "step": 4822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.676383259641908e-06,
+      "loss": 0.6411,
+      "step": 4823
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.672681995411216e-06,
+      "loss": 0.7694,
+      "step": 4824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.66898091626626e-06,
+      "loss": 0.6672,
+      "step": 4825
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.665280022723131e-06,
+      "loss": 0.5381,
+      "step": 4826
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.661579315297892e-06,
+      "loss": 0.6009,
+      "step": 4827
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.657878794506585e-06,
+      "loss": 0.7824,
+      "step": 4828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.654178460865222e-06,
+      "loss": 0.8265,
+      "step": 4829
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.650478314889786e-06,
+      "loss": 0.5495,
+      "step": 4830
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.646778357096246e-06,
+      "loss": 0.5774,
+      "step": 4831
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.64307858800053e-06,
+      "loss": 0.5459,
+      "step": 4832
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.63937900811855e-06,
+      "loss": 0.5402,
+      "step": 4833
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.635679617966185e-06,
+      "loss": 0.7333,
+      "step": 4834
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.631980418059296e-06,
+      "loss": 0.8055,
+      "step": 4835
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.628281408913702e-06,
+      "loss": 0.7715,
+      "step": 4836
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.624582591045216e-06,
+      "loss": 0.886,
+      "step": 4837
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.62088396496961e-06,
+      "loss": 0.6269,
+      "step": 4838
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.61718553120263e-06,
+      "loss": 0.5409,
+      "step": 4839
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.613487290260004e-06,
+      "loss": 0.6851,
+      "step": 4840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.609789242657421e-06,
+      "loss": 0.6031,
+      "step": 4841
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.606091388910556e-06,
+      "loss": 0.5153,
+      "step": 4842
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.602393729535042e-06,
+      "loss": 0.6539,
+      "step": 4843
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.598696265046503e-06,
+      "loss": 0.5713,
+      "step": 4844
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.594998995960516e-06,
+      "loss": 0.615,
+      "step": 4845
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.591301922792648e-06,
+      "loss": 0.6783,
+      "step": 4846
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.587605046058427e-06,
+      "loss": 0.6034,
+      "step": 4847
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.583908366273363e-06,
+      "loss": 0.7263,
+      "step": 4848
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.580211883952928e-06,
+      "loss": 0.5221,
+      "step": 4849
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.576515599612576e-06,
+      "loss": 0.4941,
+      "step": 4850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.572819513767728e-06,
+      "loss": 0.8647,
+      "step": 4851
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.56912362693378e-06,
+      "loss": 0.5497,
+      "step": 4852
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.565427939626096e-06,
+      "loss": 0.6562,
+      "step": 4853
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.56173245236002e-06,
+      "loss": 0.9416,
+      "step": 4854
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.558037165650855e-06,
+      "loss": 0.5848,
+      "step": 4855
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.554342080013894e-06,
+      "loss": 0.6237,
+      "step": 4856
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.550647195964387e-06,
+      "loss": 0.6826,
+      "step": 4857
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.546952514017563e-06,
+      "loss": 0.5043,
+      "step": 4858
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.543258034688622e-06,
+      "loss": 0.6459,
+      "step": 4859
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.53956375849273e-06,
+      "loss": 0.6217,
+      "step": 4860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.535869685945035e-06,
+      "loss": 0.4918,
+      "step": 4861
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.532175817560647e-06,
+      "loss": 0.6662,
+      "step": 4862
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.528482153854655e-06,
+      "loss": 0.6365,
+      "step": 4863
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.524788695342112e-06,
+      "loss": 0.544,
+      "step": 4864
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.52109544253805e-06,
+      "loss": 0.5458,
+      "step": 4865
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.517402395957467e-06,
+      "loss": 0.5782,
+      "step": 4866
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.513709556115335e-06,
+      "loss": 0.5328,
+      "step": 4867
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.510016923526594e-06,
+      "loss": 0.5648,
+      "step": 4868
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.50632449870616e-06,
+      "loss": 0.9277,
+      "step": 4869
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.502632282168914e-06,
+      "loss": 0.8088,
+      "step": 4870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.498940274429715e-06,
+      "loss": 0.7999,
+      "step": 4871
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.495248476003386e-06,
+      "loss": 0.5397,
+      "step": 4872
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.491556887404723e-06,
+      "loss": 0.6171,
+      "step": 4873
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.487865509148499e-06,
+      "loss": 0.5819,
+      "step": 4874
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.484174341749449e-06,
+      "loss": 0.6416,
+      "step": 4875
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.480483385722282e-06,
+      "loss": 0.5505,
+      "step": 4876
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.476792641581677e-06,
+      "loss": 0.6399,
+      "step": 4877
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.473102109842286e-06,
+      "loss": 0.5532,
+      "step": 4878
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.469411791018727e-06,
+      "loss": 0.5059,
+      "step": 4879
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.465721685625591e-06,
+      "loss": 0.5655,
+      "step": 4880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.462031794177439e-06,
+      "loss": 0.5748,
+      "step": 4881
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.458342117188804e-06,
+      "loss": 0.5861,
+      "step": 4882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.454652655174182e-06,
+      "loss": 0.6537,
+      "step": 4883
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.450963408648052e-06,
+      "loss": 0.5786,
+      "step": 4884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.447274378124845e-06,
+      "loss": 0.8736,
+      "step": 4885
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.443585564118983e-06,
+      "loss": 0.7669,
+      "step": 4886
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.439896967144838e-06,
+      "loss": 0.5514,
+      "step": 4887
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.436208587716766e-06,
+      "loss": 0.5918,
+      "step": 4888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.432520426349081e-06,
+      "loss": 0.655,
+      "step": 4889
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.42883248355608e-06,
+      "loss": 0.9054,
+      "step": 4890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.425144759852015e-06,
+      "loss": 0.5373,
+      "step": 4891
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.421457255751122e-06,
+      "loss": 0.8452,
+      "step": 4892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.417769971767596e-06,
+      "loss": 0.5632,
+      "step": 4893
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.414082908415603e-06,
+      "loss": 0.8888,
+      "step": 4894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.410396066209283e-06,
+      "loss": 0.5256,
+      "step": 4895
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.406709445662738e-06,
+      "loss": 0.9637,
+      "step": 4896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.403023047290045e-06,
+      "loss": 0.7537,
+      "step": 4897
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.399336871605247e-06,
+      "loss": 0.8409,
+      "step": 4898
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.39565091912236e-06,
+      "loss": 0.6165,
+      "step": 4899
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.391965190355359e-06,
+      "loss": 0.6342,
+      "step": 4900
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.388279685818201e-06,
+      "loss": 0.8425,
+      "step": 4901
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.384594406024802e-06,
+      "loss": 0.537,
+      "step": 4902
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.380909351489053e-06,
+      "loss": 0.6279,
+      "step": 4903
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.377224522724807e-06,
+      "loss": 0.5869,
+      "step": 4904
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.37353992024589e-06,
+      "loss": 0.6466,
+      "step": 4905
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.369855544566096e-06,
+      "loss": 0.6213,
+      "step": 4906
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.366171396199189e-06,
+      "loss": 0.703,
+      "step": 4907
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.362487475658893e-06,
+      "loss": 0.6874,
+      "step": 4908
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.358803783458915e-06,
+      "loss": 0.6402,
+      "step": 4909
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.355120320112911e-06,
+      "loss": 0.5918,
+      "step": 4910
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.351437086134523e-06,
+      "loss": 0.7827,
+      "step": 4911
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.347754082037358e-06,
+      "loss": 0.5195,
+      "step": 4912
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.344071308334976e-06,
+      "loss": 0.5348,
+      "step": 4913
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.340388765540923e-06,
+      "loss": 0.6292,
+      "step": 4914
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.336706454168701e-06,
+      "loss": 0.5897,
+      "step": 4915
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.333024374731786e-06,
+      "loss": 0.6209,
+      "step": 4916
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.32934252774362e-06,
+      "loss": 0.5027,
+      "step": 4917
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.325660913717613e-06,
+      "loss": 0.5635,
+      "step": 4918
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.321979533167137e-06,
+      "loss": 0.6605,
+      "step": 4919
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.318298386605542e-06,
+      "loss": 0.5873,
+      "step": 4920
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.314617474546136e-06,
+      "loss": 0.6971,
+      "step": 4921
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.310936797502201e-06,
+      "loss": 0.5654,
+      "step": 4922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.307256355986979e-06,
+      "loss": 0.4932,
+      "step": 4923
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.303576150513688e-06,
+      "loss": 0.6052,
+      "step": 4924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.299896181595504e-06,
+      "loss": 0.6141,
+      "step": 4925
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.296216449745578e-06,
+      "loss": 0.8947,
+      "step": 4926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.29253695547702e-06,
+      "loss": 0.5578,
+      "step": 4927
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.288857699302912e-06,
+      "loss": 0.6037,
+      "step": 4928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.285178681736305e-06,
+      "loss": 0.6135,
+      "step": 4929
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.281499903290215e-06,
+      "loss": 0.7071,
+      "step": 4930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27782136447762e-06,
+      "loss": 0.6012,
+      "step": 4931
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27414306581147e-06,
+      "loss": 0.6677,
+      "step": 4932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.270465007804676e-06,
+      "loss": 0.4942,
+      "step": 4933
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.266787190970123e-06,
+      "loss": 0.5757,
+      "step": 4934
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.263109615820656e-06,
+      "loss": 0.8832,
+      "step": 4935
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.25943228286909e-06,
+      "loss": 0.7266,
+      "step": 4936
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.255755192628203e-06,
+      "loss": 0.6119,
+      "step": 4937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.252078345610742e-06,
+      "loss": 1.0204,
+      "step": 4938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.248401742329417e-06,
+      "loss": 0.5301,
+      "step": 4939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.24472538329691e-06,
+      "loss": 0.5236,
+      "step": 4940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.241049269025862e-06,
+      "loss": 0.5102,
+      "step": 4941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.237373400028885e-06,
+      "loss": 0.4828,
+      "step": 4942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.233697776818551e-06,
+      "loss": 0.5397,
+      "step": 4943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.230022399907408e-06,
+      "loss": 0.4869,
+      "step": 4944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.226347269807956e-06,
+      "loss": 0.6255,
+      "step": 4945
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.222672387032673e-06,
+      "loss": 0.5866,
+      "step": 4946
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.218997752093988e-06,
+      "loss": 0.6679,
+      "step": 4947
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.215323365504316e-06,
+      "loss": 0.5703,
+      "step": 4948
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.211649227776021e-06,
+      "loss": 0.7195,
+      "step": 4949
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.207975339421438e-06,
+      "loss": 0.5429,
+      "step": 4950
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.204301700952866e-06,
+      "loss": 0.6201,
+      "step": 4951
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.200628312882569e-06,
+      "loss": 0.6654,
+      "step": 4952
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.196955175722778e-06,
+      "loss": 0.6589,
+      "step": 4953
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.193282289985688e-06,
+      "loss": 0.7458,
+      "step": 4954
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.189609656183458e-06,
+      "loss": 0.6081,
+      "step": 4955
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.185937274828209e-06,
+      "loss": 0.5659,
+      "step": 4956
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.182265146432036e-06,
+      "loss": 0.5739,
+      "step": 4957
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.17859327150699e-06,
+      "loss": 0.9557,
+      "step": 4958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.174921650565091e-06,
+      "loss": 0.7181,
+      "step": 4959
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.171250284118321e-06,
+      "loss": 0.7233,
+      "step": 4960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.16757917267863e-06,
+      "loss": 0.6266,
+      "step": 4961
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.163908316757927e-06,
+      "loss": 0.5679,
+      "step": 4962
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.160237716868093e-06,
+      "loss": 0.6212,
+      "step": 4963
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.156567373520964e-06,
+      "loss": 0.5005,
+      "step": 4964
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.152897287228345e-06,
+      "loss": 0.593,
+      "step": 4965
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.149227458502011e-06,
+      "loss": 0.6609,
+      "step": 4966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.145557887853692e-06,
+      "loss": 0.5394,
+      "step": 4967
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.141888575795087e-06,
+      "loss": 0.6557,
+      "step": 4968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.138219522837854e-06,
+      "loss": 0.7235,
+      "step": 4969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.134550729493622e-06,
+      "loss": 0.604,
+      "step": 4970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.130882196273978e-06,
+      "loss": 0.5757,
+      "step": 4971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.127213923690474e-06,
+      "loss": 0.8628,
+      "step": 4972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.123545912254627e-06,
+      "loss": 0.7065,
+      "step": 4973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.119878162477919e-06,
+      "loss": 0.5514,
+      "step": 4974
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11621067487179e-06,
+      "loss": 0.6038,
+      "step": 4975
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.112543449947652e-06,
+      "loss": 0.6439,
+      "step": 4976
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.108876488216867e-06,
+      "loss": 0.6353,
+      "step": 4977
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.105209790190778e-06,
+      "loss": 0.9354,
+      "step": 4978
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.101543356380676e-06,
+      "loss": 0.4784,
+      "step": 4979
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.097877187297824e-06,
+      "loss": 0.5117,
+      "step": 4980
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.094211283453441e-06,
+      "loss": 0.601,
+      "step": 4981
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.090545645358719e-06,
+      "loss": 0.5377,
+      "step": 4982
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.086880273524798e-06,
+      "loss": 0.7343,
+      "step": 4983
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.083215168462798e-06,
+      "loss": 0.757,
+      "step": 4984
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.079550330683793e-06,
+      "loss": 0.5504,
+      "step": 4985
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.075885760698818e-06,
+      "loss": 0.597,
+      "step": 4986
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.072221459018877e-06,
+      "loss": 0.5513,
+      "step": 4987
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.068557426154925e-06,
+      "loss": 0.6492,
+      "step": 4988
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.064893662617895e-06,
+      "loss": 0.5678,
+      "step": 4989
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.06123016891867e-06,
+      "loss": 0.5816,
+      "step": 4990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.057566945568103e-06,
+      "loss": 0.5733,
+      "step": 4991
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.053903993077002e-06,
+      "loss": 0.6869,
+      "step": 4992
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.050241311956148e-06,
+      "loss": 0.6841,
+      "step": 4993
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.046578902716271e-06,
+      "loss": 0.8664,
+      "step": 4994
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.042916765868074e-06,
+      "loss": 0.6147,
+      "step": 4995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.039254901922216e-06,
+      "loss": 0.5286,
+      "step": 4996
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.035593311389322e-06,
+      "loss": 0.6061,
+      "step": 4997
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.031931994779972e-06,
+      "loss": 0.7092,
+      "step": 4998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.02827095260472e-06,
+      "loss": 0.5549,
+      "step": 4999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.024610185374065e-06,
+      "loss": 0.6936,
+      "step": 5000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.020949693598484e-06,
+      "loss": 0.7125,
+      "step": 5001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.017289477788402e-06,
+      "loss": 0.4475,
+      "step": 5002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.01362953845422e-06,
+      "loss": 0.6939,
+      "step": 5003
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.009969876106287e-06,
+      "loss": 0.6065,
+      "step": 5004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.006310491254919e-06,
+      "loss": 0.5988,
+      "step": 5005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.002651384410397e-06,
+      "loss": 0.6991,
+      "step": 5006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.998992556082955e-06,
+      "loss": 0.7955,
+      "step": 5007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.995334006782793e-06,
+      "loss": 0.5841,
+      "step": 5008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.991675737020071e-06,
+      "loss": 0.5982,
+      "step": 5009
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.988017747304916e-06,
+      "loss": 0.6429,
+      "step": 5010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.984360038147403e-06,
+      "loss": 0.5123,
+      "step": 5011
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.980702610057581e-06,
+      "loss": 0.5093,
+      "step": 5012
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.97704546354545e-06,
+      "loss": 0.6311,
+      "step": 5013
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.973388599120979e-06,
+      "loss": 0.5814,
+      "step": 5014
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.969732017294089e-06,
+      "loss": 0.7779,
+      "step": 5015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.966075718574671e-06,
+      "loss": 0.5637,
+      "step": 5016
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.962419703472568e-06,
+      "loss": 0.6704,
+      "step": 5017
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.958763972497591e-06,
+      "loss": 0.5127,
+      "step": 5018
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.955108526159502e-06,
+      "loss": 0.7294,
+      "step": 5019
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.951453364968031e-06,
+      "loss": 0.8742,
+      "step": 5020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.947798489432872e-06,
+      "loss": 0.604,
+      "step": 5021
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.944143900063666e-06,
+      "loss": 0.8262,
+      "step": 5022
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.940489597370025e-06,
+      "loss": 0.6046,
+      "step": 5023
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.936835581861513e-06,
+      "loss": 0.788,
+      "step": 5024
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.933181854047666e-06,
+      "loss": 0.6213,
+      "step": 5025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.929528414437965e-06,
+      "loss": 0.7223,
+      "step": 5026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.925875263541863e-06,
+      "loss": 0.625,
+      "step": 5027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.922222401868763e-06,
+      "loss": 0.5805,
+      "step": 5028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.918569829928038e-06,
+      "loss": 0.8335,
+      "step": 5029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91491754822901e-06,
+      "loss": 0.5248,
+      "step": 5030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.91126555728097e-06,
+      "loss": 0.613,
+      "step": 5031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.907613857593159e-06,
+      "loss": 0.5939,
+      "step": 5032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.903962449674788e-06,
+      "loss": 0.6869,
+      "step": 5033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.900311334035015e-06,
+      "loss": 0.6803,
+      "step": 5034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.896660511182972e-06,
+      "loss": 0.6527,
+      "step": 5035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.893009981627735e-06,
+      "loss": 0.7451,
+      "step": 5036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.88935974587835e-06,
+      "loss": 0.8732,
+      "step": 5037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.885709804443817e-06,
+      "loss": 0.5199,
+      "step": 5038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.882060157833095e-06,
+      "loss": 0.4883,
+      "step": 5039
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.878410806555106e-06,
+      "loss": 0.7856,
+      "step": 5040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.87476175111873e-06,
+      "loss": 0.5932,
+      "step": 5041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.871112992032798e-06,
+      "loss": 0.6702,
+      "step": 5042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.86746452980611e-06,
+      "loss": 0.6926,
+      "step": 5043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.863816364947416e-06,
+      "loss": 0.5805,
+      "step": 5044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.860168497965432e-06,
+      "loss": 0.5653,
+      "step": 5045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.856520929368825e-06,
+      "loss": 0.7747,
+      "step": 5046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.85287365966623e-06,
+      "loss": 0.5843,
+      "step": 5047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.84922668936623e-06,
+      "loss": 0.8835,
+      "step": 5048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.845580018977372e-06,
+      "loss": 0.6409,
+      "step": 5049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.841933649008161e-06,
+      "loss": 0.8584,
+      "step": 5050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.838287579967059e-06,
+      "loss": 0.5625,
+      "step": 5051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.834641812362484e-06,
+      "loss": 0.6037,
+      "step": 5052
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.830996346702819e-06,
+      "loss": 0.7765,
+      "step": 5053
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.827351183496396e-06,
+      "loss": 0.6688,
+      "step": 5054
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.823706323251512e-06,
+      "loss": 0.7203,
+      "step": 5055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.820061766476414e-06,
+      "loss": 0.6778,
+      "step": 5056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.816417513679313e-06,
+      "loss": 0.6644,
+      "step": 5057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.812773565368382e-06,
+      "loss": 0.7184,
+      "step": 5058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.809129922051737e-06,
+      "loss": 0.5968,
+      "step": 5059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.805486584237467e-06,
+      "loss": 0.6327,
+      "step": 5060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.801843552433607e-06,
+      "loss": 0.5755,
+      "step": 5061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.798200827148157e-06,
+      "loss": 0.6255,
+      "step": 5062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.794558408889065e-06,
+      "loss": 0.5938,
+      "step": 5063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.790916298164251e-06,
+      "loss": 0.5981,
+      "step": 5064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.787274495481577e-06,
+      "loss": 0.6344,
+      "step": 5065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.78363300134887e-06,
+      "loss": 0.9238,
+      "step": 5066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.77999181627391e-06,
+      "loss": 0.5516,
+      "step": 5067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.776350940764441e-06,
+      "loss": 0.553,
+      "step": 5068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.772710375328156e-06,
+      "loss": 0.4942,
+      "step": 5069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.769070120472708e-06,
+      "loss": 0.647,
+      "step": 5070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.765430176705707e-06,
+      "loss": 0.693,
+      "step": 5071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.76179054453472e-06,
+      "loss": 0.5246,
+      "step": 5072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.758151224467266e-06,
+      "loss": 0.6636,
+      "step": 5073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.754512217010828e-06,
+      "loss": 0.6271,
+      "step": 5074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.750873522672842e-06,
+      "loss": 0.5821,
+      "step": 5075
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.747235141960693e-06,
+      "loss": 0.6442,
+      "step": 5076
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.74359707538174e-06,
+      "loss": 0.546,
+      "step": 5077
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.73995932344328e-06,
+      "loss": 0.5033,
+      "step": 5078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.736321886652576e-06,
+      "loss": 0.7728,
+      "step": 5079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.732684765516844e-06,
+      "loss": 0.5311,
+      "step": 5080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.729047960543257e-06,
+      "loss": 0.5842,
+      "step": 5081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.725411472238941e-06,
+      "loss": 0.8263,
+      "step": 5082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.721775301110985e-06,
+      "loss": 0.7035,
+      "step": 5083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.718139447666423e-06,
+      "loss": 0.6527,
+      "step": 5084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.714503912412256e-06,
+      "loss": 0.5964,
+      "step": 5085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.710868695855434e-06,
+      "loss": 0.5554,
+      "step": 5086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.707233798502862e-06,
+      "loss": 0.5042,
+      "step": 5087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.703599220861404e-06,
+      "loss": 0.928,
+      "step": 5088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.699964963437879e-06,
+      "loss": 0.7292,
+      "step": 5089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.696331026739057e-06,
+      "loss": 0.8859,
+      "step": 5090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.69269741127167e-06,
+      "loss": 0.7128,
+      "step": 5091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.6890641175424e-06,
+      "loss": 0.8496,
+      "step": 5092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.685431146057886e-06,
+      "loss": 0.5538,
+      "step": 5093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.681798497324717e-06,
+      "loss": 0.5564,
+      "step": 5094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.67816617184945e-06,
+      "loss": 0.9036,
+      "step": 5095
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.674534170138587e-06,
+      "loss": 0.5452,
+      "step": 5096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.670902492698584e-06,
+      "loss": 0.5743,
+      "step": 5097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.667271140035858e-06,
+      "loss": 0.6469,
+      "step": 5098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.663640112656771e-06,
+      "loss": 0.662,
+      "step": 5099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.660009411067653e-06,
+      "loss": 0.5473,
+      "step": 5100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.656379035774776e-06,
+      "loss": 0.6053,
+      "step": 5101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.652748987284375e-06,
+      "loss": 0.5808,
+      "step": 5102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.649119266102634e-06,
+      "loss": 0.6779,
+      "step": 5103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.645489872735696e-06,
+      "loss": 0.5495,
+      "step": 5104
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.641860807689652e-06,
+      "loss": 0.5274,
+      "step": 5105
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.638232071470556e-06,
+      "loss": 0.6808,
+      "step": 5106
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.63460366458441e-06,
+      "loss": 0.6097,
+      "step": 5107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.63097558753717e-06,
+      "loss": 0.6272,
+      "step": 5108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6273478408347464e-06,
+      "loss": 0.9165,
+      "step": 5109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.62372042498301e-06,
+      "loss": 0.8079,
+      "step": 5110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.620093340487772e-06,
+      "loss": 0.5618,
+      "step": 5111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6164665878548086e-06,
+      "loss": 0.5822,
+      "step": 5112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.612840167589851e-06,
+      "loss": 0.6809,
+      "step": 5113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6092140801985735e-06,
+      "loss": 0.6523,
+      "step": 5114
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.605588326186616e-06,
+      "loss": 0.6159,
+      "step": 5115
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.601962906059559e-06,
+      "loss": 0.6023,
+      "step": 5116
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.598337820322949e-06,
+      "loss": 0.7771,
+      "step": 5117
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.594713069482278e-06,
+      "loss": 0.5077,
+      "step": 5118
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.591088654042995e-06,
+      "loss": 0.6985,
+      "step": 5119
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.587464574510497e-06,
+      "loss": 0.5205,
+      "step": 5120
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.583840831390143e-06,
+      "loss": 0.6796,
+      "step": 5121
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5802174251872365e-06,
+      "loss": 0.6092,
+      "step": 5122
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.576594356407039e-06,
+      "loss": 0.8476,
+      "step": 5123
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.572971625554763e-06,
+      "loss": 0.6959,
+      "step": 5124
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.569349233135576e-06,
+      "loss": 0.7445,
+      "step": 5125
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.565727179654593e-06,
+      "loss": 0.6608,
+      "step": 5126
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.562105465616889e-06,
+      "loss": 0.6603,
+      "step": 5127
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.558484091527487e-06,
+      "loss": 0.7917,
+      "step": 5128
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.554863057891364e-06,
+      "loss": 0.8102,
+      "step": 5129
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.551242365213448e-06,
+      "loss": 0.7052,
+      "step": 5130
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.547622013998619e-06,
+      "loss": 0.839,
+      "step": 5131
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.544002004751718e-06,
+      "loss": 0.8908,
+      "step": 5132
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.540382337977524e-06,
+      "loss": 0.5185,
+      "step": 5133
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5367630141807825e-06,
+      "loss": 0.6113,
+      "step": 5134
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5331440338661775e-06,
+      "loss": 0.6656,
+      "step": 5135
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.529525397538357e-06,
+      "loss": 0.7707,
+      "step": 5136
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5259071057019125e-06,
+      "loss": 0.6597,
+      "step": 5137
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.522289158861393e-06,
+      "loss": 0.6028,
+      "step": 5138
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.518671557521297e-06,
+      "loss": 0.6314,
+      "step": 5139
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.515054302186074e-06,
+      "loss": 0.6067,
+      "step": 5140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.511437393360127e-06,
+      "loss": 0.8334,
+      "step": 5141
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.507820831547812e-06,
+      "loss": 0.6281,
+      "step": 5142
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.50420461725343e-06,
+      "loss": 0.8715,
+      "step": 5143
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5005887509812435e-06,
+      "loss": 0.6154,
+      "step": 5144
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4969732332354575e-06,
+      "loss": 0.6892,
+      "step": 5145
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.493358064520235e-06,
+      "loss": 1.0167,
+      "step": 5146
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.489743245339682e-06,
+      "loss": 0.6364,
+      "step": 5147
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.48612877619787e-06,
+      "loss": 0.7558,
+      "step": 5148
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4825146575988e-06,
+      "loss": 0.5661,
+      "step": 5149
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4789008900464514e-06,
+      "loss": 0.6626,
+      "step": 5150
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4752874740447304e-06,
+      "loss": 0.8257,
+      "step": 5151
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.471674410097511e-06,
+      "loss": 0.6313,
+      "step": 5152
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.468061698708604e-06,
+      "loss": 0.7866,
+      "step": 5153
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.464449340381786e-06,
+      "loss": 0.6104,
+      "step": 5154
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.460837335620766e-06,
+      "loss": 0.5,
+      "step": 5155
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4572256849292256e-06,
+      "loss": 0.4427,
+      "step": 5156
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.453614388810777e-06,
+      "loss": 0.6969,
+      "step": 5157
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.450003447768997e-06,
+      "loss": 0.6696,
+      "step": 5158
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.446392862307404e-06,
+      "loss": 0.8844,
+      "step": 5159
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.442782632929473e-06,
+      "loss": 0.5483,
+      "step": 5160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.439172760138625e-06,
+      "loss": 0.7659,
+      "step": 5161
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.435563244438233e-06,
+      "loss": 0.5728,
+      "step": 5162
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.431954086331618e-06,
+      "loss": 0.605,
+      "step": 5163
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.42834528632206e-06,
+      "loss": 0.9631,
+      "step": 5164
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.424736844912777e-06,
+      "loss": 0.6714,
+      "step": 5165
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.421128762606943e-06,
+      "loss": 0.615,
+      "step": 5166
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.41752103990768e-06,
+      "loss": 0.6722,
+      "step": 5167
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.413913677318061e-06,
+      "loss": 0.5075,
+      "step": 5168
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.410306675341116e-06,
+      "loss": 0.6569,
+      "step": 5169
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.406700034479809e-06,
+      "loss": 0.8308,
+      "step": 5170
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.403093755237068e-06,
+      "loss": 0.9394,
+      "step": 5171
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.399487838115761e-06,
+      "loss": 0.7265,
+      "step": 5172
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.395882283618712e-06,
+      "loss": 0.8246,
+      "step": 5173
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3922770922486896e-06,
+      "loss": 0.8235,
+      "step": 5174
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.388672264508415e-06,
+      "loss": 0.7609,
+      "step": 5175
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.385067800900557e-06,
+      "loss": 0.5319,
+      "step": 5176
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3814637019277365e-06,
+      "loss": 0.638,
+      "step": 5177
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.377859968092518e-06,
+      "loss": 0.5861,
+      "step": 5178
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.374256599897422e-06,
+      "loss": 0.6034,
+      "step": 5179
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.370653597844909e-06,
+      "loss": 0.6189,
+      "step": 5180
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.367050962437401e-06,
+      "loss": 0.5822,
+      "step": 5181
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3634486941772555e-06,
+      "loss": 0.6039,
+      "step": 5182
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.35984679356679e-06,
+      "loss": 0.5086,
+      "step": 5183
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.356245261108259e-06,
+      "loss": 0.5352,
+      "step": 5184
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.352644097303881e-06,
+      "loss": 0.6362,
+      "step": 5185
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.349043302655805e-06,
+      "loss": 0.6204,
+      "step": 5186
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.345442877666146e-06,
+      "loss": 0.5746,
+      "step": 5187
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.341842822836959e-06,
+      "loss": 0.6076,
+      "step": 5188
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.338243138670243e-06,
+      "loss": 0.5682,
+      "step": 5189
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.334643825667955e-06,
+      "loss": 0.6414,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.331044884331992e-06,
+      "loss": 0.9955,
+      "step": 5191
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.327446315164206e-06,
+      "loss": 0.6596,
+      "step": 5192
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.323848118666388e-06,
+      "loss": 0.5445,
+      "step": 5193
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.32025029534029e-06,
+      "loss": 0.7026,
+      "step": 5194
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3166528456876e-06,
+      "loss": 0.5567,
+      "step": 5195
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.313055770209961e-06,
+      "loss": 0.6857,
+      "step": 5196
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3094590694089565e-06,
+      "loss": 0.8016,
+      "step": 5197
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3058627437861294e-06,
+      "loss": 0.5121,
+      "step": 5198
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.302266793842958e-06,
+      "loss": 0.545,
+      "step": 5199
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.298671220080879e-06,
+      "loss": 0.6326,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.295076023001268e-06,
+      "loss": 0.5363,
+      "step": 5201
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.291481203105452e-06,
+      "loss": 0.6106,
+      "step": 5202
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.287886760894704e-06,
+      "loss": 0.8901,
+      "step": 5203
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2842926968702455e-06,
+      "loss": 0.6016,
+      "step": 5204
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.280699011533247e-06,
+      "loss": 0.5971,
+      "step": 5205
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.277105705384824e-06,
+      "loss": 0.5254,
+      "step": 5206
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.273512778926039e-06,
+      "loss": 0.6137,
+      "step": 5207
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.269920232657901e-06,
+      "loss": 0.9575,
+      "step": 5208
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.266328067081367e-06,
+      "loss": 0.6251,
+      "step": 5209
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.26273628269734e-06,
+      "loss": 0.8095,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2591448800066745e-06,
+      "loss": 0.5231,
+      "step": 5211
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.25555385951016e-06,
+      "loss": 0.9599,
+      "step": 5212
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.251963221708551e-06,
+      "loss": 0.5617,
+      "step": 5213
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.24837296710253e-06,
+      "loss": 0.5491,
+      "step": 5214
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.244783096192738e-06,
+      "loss": 0.6058,
+      "step": 5215
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.241193609479756e-06,
+      "loss": 0.7334,
+      "step": 5216
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2376045074641185e-06,
+      "loss": 0.5456,
+      "step": 5217
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.234015790646298e-06,
+      "loss": 0.8086,
+      "step": 5218
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.230427459526721e-06,
+      "loss": 0.7216,
+      "step": 5219
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.226839514605754e-06,
+      "loss": 0.83,
+      "step": 5220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.223251956383713e-06,
+      "loss": 0.7324,
+      "step": 5221
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.219664785360858e-06,
+      "loss": 0.7566,
+      "step": 5222
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2160780020373955e-06,
+      "loss": 0.811,
+      "step": 5223
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2124916069134855e-06,
+      "loss": 0.5041,
+      "step": 5224
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.20890560048922e-06,
+      "loss": 0.7344,
+      "step": 5225
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.205319983264648e-06,
+      "loss": 0.7986,
+      "step": 5226
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.201734755739758e-06,
+      "loss": 0.5985,
+      "step": 5227
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.198149918414488e-06,
+      "loss": 0.6113,
+      "step": 5228
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.194565471788715e-06,
+      "loss": 0.5726,
+      "step": 5229
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.190981416362275e-06,
+      "loss": 0.5744,
+      "step": 5230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.187397752634933e-06,
+      "loss": 0.5308,
+      "step": 5231
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.183814481106411e-06,
+      "loss": 0.5927,
+      "step": 5232
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.180231602276372e-06,
+      "loss": 0.9529,
+      "step": 5233
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.176649116644425e-06,
+      "loss": 0.6079,
+      "step": 5234
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.173067024710123e-06,
+      "loss": 0.5155,
+      "step": 5235
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.169485326972968e-06,
+      "loss": 0.5828,
+      "step": 5236
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1659040239324e-06,
+      "loss": 0.5581,
+      "step": 5237
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.162323116087811e-06,
+      "loss": 0.5672,
+      "step": 5238
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.158742603938531e-06,
+      "loss": 0.6866,
+      "step": 5239
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1551624879838445e-06,
+      "loss": 0.6332,
+      "step": 5240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1515827687229685e-06,
+      "loss": 0.5487,
+      "step": 5241
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1480034466550775e-06,
+      "loss": 0.6729,
+      "step": 5242
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.144424522279283e-06,
+      "loss": 0.9815,
+      "step": 5243
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.14084599609464e-06,
+      "loss": 0.9258,
+      "step": 5244
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.137267868600153e-06,
+      "loss": 0.5944,
+      "step": 5245
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.133690140294764e-06,
+      "loss": 0.9531,
+      "step": 5246
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.13011281167737e-06,
+      "loss": 0.6526,
+      "step": 5247
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.126535883246799e-06,
+      "loss": 0.7826,
+      "step": 5248
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.122959355501836e-06,
+      "loss": 0.7062,
+      "step": 5249
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.119383228941199e-06,
+      "loss": 0.6476,
+      "step": 5250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.115807504063561e-06,
+      "loss": 0.5171,
+      "step": 5251
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.112232181367527e-06,
+      "loss": 0.6358,
+      "step": 5252
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.108657261351657e-06,
+      "loss": 0.6361,
+      "step": 5253
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.105082744514446e-06,
+      "loss": 0.6911,
+      "step": 5254
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.101508631354341e-06,
+      "loss": 0.5494,
+      "step": 5255
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0979349223697245e-06,
+      "loss": 0.6431,
+      "step": 5256
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.094361618058931e-06,
+      "loss": 0.9594,
+      "step": 5257
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.090788718920229e-06,
+      "loss": 0.7845,
+      "step": 5258
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.087216225451841e-06,
+      "loss": 0.9506,
+      "step": 5259
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.08364413815192e-06,
+      "loss": 0.8188,
+      "step": 5260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.080072457518579e-06,
+      "loss": 0.6315,
+      "step": 5261
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.07650118404986e-06,
+      "loss": 0.9394,
+      "step": 5262
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.072930318243758e-06,
+      "loss": 0.6683,
+      "step": 5263
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.069359860598199e-06,
+      "loss": 0.9175,
+      "step": 5264
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0657898116110705e-06,
+      "loss": 0.59,
+      "step": 5265
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.062220171780181e-06,
+      "loss": 0.5312,
+      "step": 5266
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.0586509416033e-06,
+      "loss": 0.6002,
+      "step": 5267
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.055082121578131e-06,
+      "loss": 0.6022,
+      "step": 5268
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.051513712202325e-06,
+      "loss": 0.5268,
+      "step": 5269
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.047945713973468e-06,
+      "loss": 0.9,
+      "step": 5270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.044378127389098e-06,
+      "loss": 0.5865,
+      "step": 5271
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.040810952946688e-06,
+      "loss": 0.825,
+      "step": 5272
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.037244191143662e-06,
+      "loss": 0.7995,
+      "step": 5273
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.033677842477377e-06,
+      "loss": 0.7476,
+      "step": 5274
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.030111907445139e-06,
+      "loss": 0.7186,
+      "step": 5275
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0265463865441915e-06,
+      "loss": 0.6638,
+      "step": 5276
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0229812802717275e-06,
+      "loss": 0.5234,
+      "step": 5277
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.01941658912487e-06,
+      "loss": 0.9524,
+      "step": 5278
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.015852313600698e-06,
+      "loss": 0.5696,
+      "step": 5279
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0122884541962275e-06,
+      "loss": 0.8254,
+      "step": 5280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.00872501140841e-06,
+      "loss": 0.7284,
+      "step": 5281
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.005161985734148e-06,
+      "loss": 0.5554,
+      "step": 5282
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.001599377670278e-06,
+      "loss": 0.8914,
+      "step": 5283
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.998037187713588e-06,
+      "loss": 0.6821,
+      "step": 5284
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.994475416360796e-06,
+      "loss": 0.5642,
+      "step": 5285
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.990914064108569e-06,
+      "loss": 0.6391,
+      "step": 5286
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9873531314535156e-06,
+      "loss": 0.6956,
+      "step": 5287
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.983792618892184e-06,
+      "loss": 0.9488,
+      "step": 5288
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.980232526921062e-06,
+      "loss": 0.4553,
+      "step": 5289
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.976672856036586e-06,
+      "loss": 0.6997,
+      "step": 5290
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.973113606735123e-06,
+      "loss": 0.7237,
+      "step": 5291
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.96955477951299e-06,
+      "loss": 0.758,
+      "step": 5292
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.96599637486644e-06,
+      "loss": 0.7209,
+      "step": 5293
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.962438393291673e-06,
+      "loss": 0.6213,
+      "step": 5294
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9588808352848195e-06,
+      "loss": 0.4986,
+      "step": 5295
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9553237013419595e-06,
+      "loss": 0.5237,
+      "step": 5296
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.951766991959118e-06,
+      "loss": 0.5629,
+      "step": 5297
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.948210707632248e-06,
+      "loss": 0.687,
+      "step": 5298
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.944654848857254e-06,
+      "loss": 0.6199,
+      "step": 5299
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.941099416129973e-06,
+      "loss": 0.4976,
+      "step": 5300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9375444099461906e-06,
+      "loss": 0.6426,
+      "step": 5301
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.933989830801627e-06,
+      "loss": 0.6545,
+      "step": 5302
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9304356791919445e-06,
+      "loss": 0.8814,
+      "step": 5303
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.926881955612746e-06,
+      "loss": 0.6967,
+      "step": 5304
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.923328660559576e-06,
+      "loss": 0.5495,
+      "step": 5305
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.919775794527915e-06,
+      "loss": 0.6044,
+      "step": 5306
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.916223358013192e-06,
+      "loss": 0.6092,
+      "step": 5307
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.912671351510766e-06,
+      "loss": 0.5223,
+      "step": 5308
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.909119775515945e-06,
+      "loss": 0.6423,
+      "step": 5309
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9055686305239665e-06,
+      "loss": 0.6185,
+      "step": 5310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.902017917030022e-06,
+      "loss": 0.7906,
+      "step": 5311
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.898467635529229e-06,
+      "loss": 0.7286,
+      "step": 5312
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.894917786516655e-06,
+      "loss": 0.6353,
+      "step": 5313
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.891368370487299e-06,
+      "loss": 0.8658,
+      "step": 5314
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.887819387936104e-06,
+      "loss": 0.6672,
+      "step": 5315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.884270839357958e-06,
+      "loss": 0.6288,
+      "step": 5316
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.880722725247677e-06,
+      "loss": 0.7744,
+      "step": 5317
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8771750461000264e-06,
+      "loss": 0.7915,
+      "step": 5318
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8736278024097e-06,
+      "loss": 0.4476,
+      "step": 5319
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.870080994671344e-06,
+      "loss": 0.4968,
+      "step": 5320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.866534623379535e-06,
+      "loss": 0.7294,
+      "step": 5321
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.862988689028791e-06,
+      "loss": 0.5787,
+      "step": 5322
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8594431921135675e-06,
+      "loss": 0.7021,
+      "step": 5323
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8558981331282634e-06,
+      "loss": 0.4908,
+      "step": 5324
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.852353512567211e-06,
+      "loss": 0.6805,
+      "step": 5325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.848809330924688e-06,
+      "loss": 0.5673,
+      "step": 5326
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.845265588694904e-06,
+      "loss": 0.7382,
+      "step": 5327
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8417222863720124e-06,
+      "loss": 0.6191,
+      "step": 5328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8381794244501e-06,
+      "loss": 0.5029,
+      "step": 5329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.834637003423202e-06,
+      "loss": 0.7101,
+      "step": 5330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.831095023785277e-06,
+      "loss": 0.682,
+      "step": 5331
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.827553486030239e-06,
+      "loss": 0.7497,
+      "step": 5332
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.824012390651922e-06,
+      "loss": 0.7959,
+      "step": 5333
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8204717381441175e-06,
+      "loss": 0.6711,
+      "step": 5334
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.816931529000546e-06,
+      "loss": 0.5782,
+      "step": 5335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.813391763714861e-06,
+      "loss": 0.76,
+      "step": 5336
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.809852442780664e-06,
+      "loss": 0.5782,
+      "step": 5337
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.806313566691486e-06,
+      "loss": 0.8281,
+      "step": 5338
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.802775135940803e-06,
+      "loss": 0.533,
+      "step": 5339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7992371510220235e-06,
+      "loss": 0.5339,
+      "step": 5340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7956996124284994e-06,
+      "loss": 0.7632,
+      "step": 5341
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.7921625206535125e-06,
+      "loss": 0.6723,
+      "step": 5342
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.788625876190291e-06,
+      "loss": 0.6366,
+      "step": 5343
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.785089679531994e-06,
+      "loss": 0.4954,
+      "step": 5344
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.781553931171723e-06,
+      "loss": 0.8187,
+      "step": 5345
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.778018631602512e-06,
+      "loss": 0.5661,
+      "step": 5346
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.774483781317338e-06,
+      "loss": 0.7825,
+      "step": 5347
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.770949380809111e-06,
+      "loss": 0.5549,
+      "step": 5348
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.767415430570681e-06,
+      "loss": 0.5278,
+      "step": 5349
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.763881931094831e-06,
+      "loss": 0.5746,
+      "step": 5350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.760348882874288e-06,
+      "loss": 0.5548,
+      "step": 5351
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.756816286401707e-06,
+      "loss": 0.5477,
+      "step": 5352
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.753284142169692e-06,
+      "loss": 0.8854,
+      "step": 5353
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.749752450670775e-06,
+      "loss": 0.547,
+      "step": 5354
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.746221212397422e-06,
+      "loss": 0.7025,
+      "step": 5355
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.742690427842049e-06,
+      "loss": 0.5809,
+      "step": 5356
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.739160097496995e-06,
+      "loss": 0.5723,
+      "step": 5357
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.735630221854544e-06,
+      "loss": 0.5536,
+      "step": 5358
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.732100801406911e-06,
+      "loss": 0.8979,
+      "step": 5359
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.728571836646255e-06,
+      "loss": 0.967,
+      "step": 5360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.725043328064661e-06,
+      "loss": 0.4758,
+      "step": 5361
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7215152761541616e-06,
+      "loss": 0.4457,
+      "step": 5362
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.717987681406714e-06,
+      "loss": 0.5069,
+      "step": 5363
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.714460544314227e-06,
+      "loss": 0.9551,
+      "step": 5364
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.710933865368527e-06,
+      "loss": 0.7151,
+      "step": 5365
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.707407645061395e-06,
+      "loss": 0.5409,
+      "step": 5366
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.703881883884531e-06,
+      "loss": 0.4962,
+      "step": 5367
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.700356582329586e-06,
+      "loss": 0.5395,
+      "step": 5368
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.696831740888133e-06,
+      "loss": 0.5926,
+      "step": 5369
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6933073600516906e-06,
+      "loss": 0.647,
+      "step": 5370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.689783440311715e-06,
+      "loss": 0.5689,
+      "step": 5371
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.686259982159591e-06,
+      "loss": 0.5495,
+      "step": 5372
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6827369860866385e-06,
+      "loss": 0.5967,
+      "step": 5373
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6792144525841205e-06,
+      "loss": 0.6391,
+      "step": 5374
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.675692382143227e-06,
+      "loss": 0.5501,
+      "step": 5375
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.672170775255091e-06,
+      "loss": 0.6451,
+      "step": 5376
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.668649632410775e-06,
+      "loss": 0.6001,
+      "step": 5377
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.665128954101281e-06,
+      "loss": 0.6654,
+      "step": 5378
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.661608740817541e-06,
+      "loss": 0.7776,
+      "step": 5379
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.65808899305043e-06,
+      "loss": 0.9715,
+      "step": 5380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.654569711290749e-06,
+      "loss": 0.7935,
+      "step": 5381
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.651050896029243e-06,
+      "loss": 0.6593,
+      "step": 5382
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.647532547756583e-06,
+      "loss": 0.5426,
+      "step": 5383
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6440146669633855e-06,
+      "loss": 0.6324,
+      "step": 5384
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.640497254140188e-06,
+      "loss": 0.6236,
+      "step": 5385
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.636980309777478e-06,
+      "loss": 0.6836,
+      "step": 5386
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.633463834365664e-06,
+      "loss": 0.5834,
+      "step": 5387
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.629947828395096e-06,
+      "loss": 0.6272,
+      "step": 5388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.626432292356064e-06,
+      "loss": 0.5802,
+      "step": 5389
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.622917226738781e-06,
+      "loss": 0.6484,
+      "step": 5390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.619402632033403e-06,
+      "loss": 0.602,
+      "step": 5391
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.615888508730013e-06,
+      "loss": 0.7144,
+      "step": 5392
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.612374857318635e-06,
+      "loss": 0.5924,
+      "step": 5393
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.608861678289223e-06,
+      "loss": 0.5454,
+      "step": 5394
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.605348972131669e-06,
+      "loss": 0.8402,
+      "step": 5395
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.601836739335793e-06,
+      "loss": 0.6023,
+      "step": 5396
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.598324980391358e-06,
+      "loss": 0.7905,
+      "step": 5397
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.594813695788048e-06,
+      "loss": 0.5913,
+      "step": 5398
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.591302886015496e-06,
+      "loss": 0.7749,
+      "step": 5399
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.587792551563257e-06,
+      "loss": 0.6583,
+      "step": 5400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.584282692920825e-06,
+      "loss": 0.5894,
+      "step": 5401
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.580773310577626e-06,
+      "loss": 0.8673,
+      "step": 5402
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.577264405023022e-06,
+      "loss": 0.5567,
+      "step": 5403
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5737559767463055e-06,
+      "loss": 0.6407,
+      "step": 5404
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.570248026236704e-06,
+      "loss": 0.5488,
+      "step": 5405
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5667405539833775e-06,
+      "loss": 0.8536,
+      "step": 5406
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.563233560475416e-06,
+      "loss": 0.5324,
+      "step": 5407
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.559727046201856e-06,
+      "loss": 0.9456,
+      "step": 5408
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.55622101165165e-06,
+      "loss": 0.5974,
+      "step": 5409
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5527154573136966e-06,
+      "loss": 0.6119,
+      "step": 5410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.549210383676817e-06,
+      "loss": 0.7323,
+      "step": 5411
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.545705791229775e-06,
+      "loss": 0.5876,
+      "step": 5412
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.542201680461259e-06,
+      "loss": 0.6092,
+      "step": 5413
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.538698051859899e-06,
+      "loss": 0.6019,
+      "step": 5414
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.535194905914247e-06,
+      "loss": 0.8732,
+      "step": 5415
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.531692243112799e-06,
+      "loss": 0.9122,
+      "step": 5416
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.528190063943973e-06,
+      "loss": 0.6035,
+      "step": 5417
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.52468836889613e-06,
+      "loss": 0.6407,
+      "step": 5418
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.521187158457555e-06,
+      "loss": 0.5034,
+      "step": 5419
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.51768643311647e-06,
+      "loss": 0.6436,
+      "step": 5420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.514186193361027e-06,
+      "loss": 0.5121,
+      "step": 5421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.510686439679313e-06,
+      "loss": 0.5184,
+      "step": 5422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.507187172559341e-06,
+      "loss": 0.703,
+      "step": 5423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5036883924890695e-06,
+      "loss": 0.8591,
+      "step": 5424
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.500190099956369e-06,
+      "loss": 0.6228,
+      "step": 5425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.496692295449061e-06,
+      "loss": 0.5719,
+      "step": 5426
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.493194979454891e-06,
+      "loss": 0.5834,
+      "step": 5427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.489698152461535e-06,
+      "loss": 0.6622,
+      "step": 5428
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.486201814956602e-06,
+      "loss": 0.7019,
+      "step": 5429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.482705967427634e-06,
+      "loss": 0.7475,
+      "step": 5430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.479210610362103e-06,
+      "loss": 0.4949,
+      "step": 5431
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.475715744247415e-06,
+      "loss": 0.4706,
+      "step": 5432
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.472221369570906e-06,
+      "loss": 0.5371,
+      "step": 5433
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.46872748681984e-06,
+      "loss": 0.6114,
+      "step": 5434
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.465234096481421e-06,
+      "loss": 0.6379,
+      "step": 5435
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.461741199042775e-06,
+      "loss": 0.7574,
+      "step": 5436
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.458248794990967e-06,
+      "loss": 0.5416,
+      "step": 5437
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.454756884812985e-06,
+      "loss": 0.5682,
+      "step": 5438
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.451265468995758e-06,
+      "loss": 0.7077,
+      "step": 5439
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.447774548026138e-06,
+      "loss": 0.6137,
+      "step": 5440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.444284122390912e-06,
+      "loss": 0.9456,
+      "step": 5441
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.440794192576793e-06,
+      "loss": 0.6241,
+      "step": 5442
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.437304759070438e-06,
+      "loss": 0.488,
+      "step": 5443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.433815822358411e-06,
+      "loss": 0.6172,
+      "step": 5444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.430327382927235e-06,
+      "loss": 0.5702,
+      "step": 5445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.426839441263345e-06,
+      "loss": 0.5339,
+      "step": 5446
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.423351997853109e-06,
+      "loss": 0.5917,
+      "step": 5447
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.419865053182832e-06,
+      "loss": 0.6839,
+      "step": 5448
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.416378607738742e-06,
+      "loss": 0.6751,
+      "step": 5449
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.412892662007003e-06,
+      "loss": 0.655,
+      "step": 5450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.409407216473704e-06,
+      "loss": 0.6489,
+      "step": 5451
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.405922271624874e-06,
+      "loss": 0.5533,
+      "step": 5452
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.402437827946457e-06,
+      "loss": 0.6383,
+      "step": 5453
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.398953885924343e-06,
+      "loss": 0.5588,
+      "step": 5454
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3954704460443385e-06,
+      "loss": 0.5253,
+      "step": 5455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.391987508792193e-06,
+      "loss": 0.6321,
+      "step": 5456
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.388505074653572e-06,
+      "loss": 0.5679,
+      "step": 5457
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3850231441140834e-06,
+      "loss": 0.8417,
+      "step": 5458
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3815417176592545e-06,
+      "loss": 0.6689,
+      "step": 5459
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.378060795774554e-06,
+      "loss": 0.7076,
+      "step": 5460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.374580378945366e-06,
+      "loss": 0.7368,
+      "step": 5461
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.371100467657015e-06,
+      "loss": 0.7971,
+      "step": 5462
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.367621062394754e-06,
+      "loss": 0.543,
+      "step": 5463
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.364142163643761e-06,
+      "loss": 0.5801,
+      "step": 5464
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3606637718891465e-06,
+      "loss": 0.6369,
+      "step": 5465
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3571858876159455e-06,
+      "loss": 0.6793,
+      "step": 5466
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.353708511309131e-06,
+      "loss": 0.6131,
+      "step": 5467
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3502316434535975e-06,
+      "loss": 0.5956,
+      "step": 5468
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.346755284534172e-06,
+      "loss": 0.543,
+      "step": 5469
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3432794350356075e-06,
+      "loss": 0.5554,
+      "step": 5470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.339804095442594e-06,
+      "loss": 0.7583,
+      "step": 5471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.336329266239738e-06,
+      "loss": 0.7413,
+      "step": 5472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.332854947911586e-06,
+      "loss": 0.6103,
+      "step": 5473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.329381140942605e-06,
+      "loss": 0.5916,
+      "step": 5474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3259078458172e-06,
+      "loss": 0.6358,
+      "step": 5475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3224350630196914e-06,
+      "loss": 0.5723,
+      "step": 5476
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.318962793034345e-06,
+      "loss": 0.6568,
+      "step": 5477
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.315491036345338e-06,
+      "loss": 0.5638,
+      "step": 5478
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.312019793436789e-06,
+      "loss": 0.5944,
+      "step": 5479
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.308549064792735e-06,
+      "loss": 0.7295,
+      "step": 5480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.305078850897149e-06,
+      "loss": 0.6068,
+      "step": 5481
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.301609152233932e-06,
+      "loss": 0.8659,
+      "step": 5482
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.298139969286907e-06,
+      "loss": 0.6548,
+      "step": 5483
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.294671302539832e-06,
+      "loss": 0.5741,
+      "step": 5484
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.291203152476385e-06,
+      "loss": 0.8436,
+      "step": 5485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.287735519580182e-06,
+      "loss": 0.6791,
+      "step": 5486
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.284268404334755e-06,
+      "loss": 0.8475,
+      "step": 5487
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.280801807223577e-06,
+      "loss": 0.573,
+      "step": 5488
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.277335728730038e-06,
+      "loss": 0.6244,
+      "step": 5489
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.273870169337463e-06,
+      "loss": 0.661,
+      "step": 5490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.270405129529096e-06,
+      "loss": 0.7512,
+      "step": 5491
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.266940609788119e-06,
+      "loss": 0.5811,
+      "step": 5492
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.263476610597634e-06,
+      "loss": 0.8257,
+      "step": 5493
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.260013132440674e-06,
+      "loss": 0.8407,
+      "step": 5494
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.256550175800197e-06,
+      "loss": 0.6478,
+      "step": 5495
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.25308774115909e-06,
+      "loss": 0.6751,
+      "step": 5496
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2496258290001675e-06,
+      "loss": 0.4777,
+      "step": 5497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.246164439806169e-06,
+      "loss": 0.8453,
+      "step": 5498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.242703574059758e-06,
+      "loss": 0.4523,
+      "step": 5499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.23924323224354e-06,
+      "loss": 0.5268,
+      "step": 5500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2357834148400295e-06,
+      "loss": 0.6171,
+      "step": 5501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.232324122331679e-06,
+      "loss": 0.5411,
+      "step": 5502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.228865355200859e-06,
+      "loss": 0.6081,
+      "step": 5503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2254071139298774e-06,
+      "loss": 0.6567,
+      "step": 5504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.221949399000957e-06,
+      "loss": 0.6119,
+      "step": 5505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.21849221089626e-06,
+      "loss": 0.6424,
+      "step": 5506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.215035550097862e-06,
+      "loss": 0.6388,
+      "step": 5507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.211579417087777e-06,
+      "loss": 0.5461,
+      "step": 5508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.208123812347936e-06,
+      "loss": 0.6326,
+      "step": 5509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2046687363602035e-06,
+      "loss": 0.5204,
+      "step": 5510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2012141896063635e-06,
+      "loss": 0.6646,
+      "step": 5511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.197760172568132e-06,
+      "loss": 0.6354,
+      "step": 5512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.194306685727149e-06,
+      "loss": 0.6537,
+      "step": 5513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.190853729564981e-06,
+      "loss": 0.6892,
+      "step": 5514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.187401304563117e-06,
+      "loss": 0.6872,
+      "step": 5515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.183949411202976e-06,
+      "loss": 0.5194,
+      "step": 5516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.180498049965902e-06,
+      "loss": 0.6443,
+      "step": 5517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1770472213331656e-06,
+      "loss": 0.6142,
+      "step": 5518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.173596925785963e-06,
+      "loss": 0.8249,
+      "step": 5519
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.170147163805414e-06,
+      "loss": 0.5998,
+      "step": 5520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.166697935872565e-06,
+      "loss": 0.7371,
+      "step": 5521
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.163249242468386e-06,
+      "loss": 0.5056,
+      "step": 5522
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.159801084073777e-06,
+      "loss": 0.6449,
+      "step": 5523
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.15635346116956e-06,
+      "loss": 0.586,
+      "step": 5524
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1529063742364844e-06,
+      "loss": 0.7342,
+      "step": 5525
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.149459823755223e-06,
+      "loss": 0.8903,
+      "step": 5526
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1460138102063745e-06,
+      "loss": 0.6272,
+      "step": 5527
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.142568334070462e-06,
+      "loss": 0.7208,
+      "step": 5528
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.139123395827936e-06,
+      "loss": 0.9678,
+      "step": 5529
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.135678995959168e-06,
+      "loss": 0.7251,
+      "step": 5530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.13223513494446e-06,
+      "loss": 0.6194,
+      "step": 5531
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.128791813264032e-06,
+      "loss": 0.5915,
+      "step": 5532
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.125349031398035e-06,
+      "loss": 0.6063,
+      "step": 5533
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.121906789826543e-06,
+      "loss": 0.5701,
+      "step": 5534
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.118465089029553e-06,
+      "loss": 0.6561,
+      "step": 5535
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.115023929486983e-06,
+      "loss": 0.5392,
+      "step": 5536
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1115833116786845e-06,
+      "loss": 0.6193,
+      "step": 5537
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.108143236084432e-06,
+      "loss": 0.6054,
+      "step": 5538
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.104703703183915e-06,
+      "loss": 0.6754,
+      "step": 5539
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.10126471345676e-06,
+      "loss": 0.5825,
+      "step": 5540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.097826267382504e-06,
+      "loss": 0.5298,
+      "step": 5541
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.094388365440622e-06,
+      "loss": 0.5782,
+      "step": 5542
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.090951008110501e-06,
+      "loss": 0.5639,
+      "step": 5543
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.087514195871465e-06,
+      "loss": 0.6233,
+      "step": 5544
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.084077929202748e-06,
+      "loss": 0.5702,
+      "step": 5545
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.080642208583521e-06,
+      "loss": 0.5725,
+      "step": 5546
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.077207034492865e-06,
+      "loss": 0.522,
+      "step": 5547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0737724074097985e-06,
+      "loss": 0.4822,
+      "step": 5548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.070338327813254e-06,
+      "loss": 0.5954,
+      "step": 5549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.066904796182094e-06,
+      "loss": 0.5776,
+      "step": 5550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.063471812995099e-06,
+      "loss": 0.5064,
+      "step": 5551
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.06003937873098e-06,
+      "loss": 0.9213,
+      "step": 5552
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.056607493868362e-06,
+      "loss": 0.582,
+      "step": 5553
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.053176158885798e-06,
+      "loss": 0.5732,
+      "step": 5554
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.049745374261775e-06,
+      "loss": 0.665,
+      "step": 5555
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0463151404746835e-06,
+      "loss": 0.7579,
+      "step": 5556
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.042885458002852e-06,
+      "loss": 0.7377,
+      "step": 5557
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.039456327324523e-06,
+      "loss": 0.5991,
+      "step": 5558
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.036027748917871e-06,
+      "loss": 0.7067,
+      "step": 5559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.032599723260983e-06,
+      "loss": 0.7367,
+      "step": 5560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.029172250831881e-06,
+      "loss": 0.693,
+      "step": 5561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.025745332108497e-06,
+      "loss": 0.5795,
+      "step": 5562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.022318967568697e-06,
+      "loss": 0.5082,
+      "step": 5563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.018893157690261e-06,
+      "loss": 0.528,
+      "step": 5564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0154679029509e-06,
+      "loss": 0.5772,
+      "step": 5565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.012043203828238e-06,
+      "loss": 0.5474,
+      "step": 5566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.008619060799832e-06,
+      "loss": 0.6673,
+      "step": 5567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.005195474343152e-06,
+      "loss": 0.8534,
+      "step": 5568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.001772444935597e-06,
+      "loss": 0.5923,
+      "step": 5569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.998349973054485e-06,
+      "loss": 0.8132,
+      "step": 5570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.994928059177056e-06,
+      "loss": 0.5801,
+      "step": 5571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.991506703780475e-06,
+      "loss": 0.5786,
+      "step": 5572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.988085907341824e-06,
+      "loss": 0.5709,
+      "step": 5573
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.984665670338118e-06,
+      "loss": 0.566,
+      "step": 5574
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.981245993246279e-06,
+      "loss": 0.5348,
+      "step": 5575
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9778268765431635e-06,
+      "loss": 0.574,
+      "step": 5576
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.97440832070554e-06,
+      "loss": 0.9262,
+      "step": 5577
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.970990326210109e-06,
+      "loss": 0.5868,
+      "step": 5578
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.967572893533482e-06,
+      "loss": 0.5378,
+      "step": 5579
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9641560231522015e-06,
+      "loss": 0.6741,
+      "step": 5580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.960739715542724e-06,
+      "loss": 0.7883,
+      "step": 5581
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.957323971181435e-06,
+      "loss": 0.7563,
+      "step": 5582
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.953908790544631e-06,
+      "loss": 0.8839,
+      "step": 5583
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.950494174108545e-06,
+      "loss": 0.5379,
+      "step": 5584
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.947080122349315e-06,
+      "loss": 0.6106,
+      "step": 5585
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.943666635743015e-06,
+      "loss": 0.5103,
+      "step": 5586
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.940253714765627e-06,
+      "loss": 0.6403,
+      "step": 5587
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.936841359893065e-06,
+      "loss": 0.9457,
+      "step": 5588
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.933429571601155e-06,
+      "loss": 0.6047,
+      "step": 5589
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.930018350365654e-06,
+      "loss": 0.7914,
+      "step": 5590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.926607696662224e-06,
+      "loss": 0.6685,
+      "step": 5591
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.923197610966469e-06,
+      "loss": 0.7084,
+      "step": 5592
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.919788093753901e-06,
+      "loss": 0.9916,
+      "step": 5593
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.91637914549995e-06,
+      "loss": 0.4413,
+      "step": 5594
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.912970766679976e-06,
+      "loss": 0.6972,
+      "step": 5595
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.909562957769251e-06,
+      "loss": 0.8625,
+      "step": 5596
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.906155719242977e-06,
+      "loss": 0.6395,
+      "step": 5597
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.902749051576262e-06,
+      "loss": 0.8821,
+      "step": 5598
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.899342955244153e-06,
+      "loss": 0.7444,
+      "step": 5599
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.895937430721601e-06,
+      "loss": 0.5624,
+      "step": 5600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.892532478483488e-06,
+      "loss": 0.5806,
+      "step": 5601
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.889128099004608e-06,
+      "loss": 0.6408,
+      "step": 5602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.885724292759683e-06,
+      "loss": 0.6613,
+      "step": 5603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.882321060223349e-06,
+      "loss": 0.6301,
+      "step": 5604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.878918401870167e-06,
+      "loss": 0.6719,
+      "step": 5605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.875516318174611e-06,
+      "loss": 0.5278,
+      "step": 5606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.872114809611085e-06,
+      "loss": 0.4962,
+      "step": 5607
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8687138766538994e-06,
+      "loss": 0.5399,
+      "step": 5608
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.865313519777295e-06,
+      "loss": 0.5195,
+      "step": 5609
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8619137394554315e-06,
+      "loss": 0.6057,
+      "step": 5610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.858514536162387e-06,
+      "loss": 0.826,
+      "step": 5611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.855115910372154e-06,
+      "loss": 0.6991,
+      "step": 5612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8517178625586504e-06,
+      "loss": 0.592,
+      "step": 5613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.848320393195708e-06,
+      "loss": 0.5387,
+      "step": 5614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.844923502757089e-06,
+      "loss": 0.5476,
+      "step": 5615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8415271917164585e-06,
+      "loss": 0.5316,
+      "step": 5616
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.838131460547417e-06,
+      "loss": 0.5615,
+      "step": 5617
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8347363097234726e-06,
+      "loss": 0.6296,
+      "step": 5618
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.831341739718055e-06,
+      "loss": 0.5553,
+      "step": 5619
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.827947751004524e-06,
+      "loss": 0.6914,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.824554344056137e-06,
+      "loss": 0.6453,
+      "step": 5621
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.821161519346088e-06,
+      "loss": 0.9407,
+      "step": 5622
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.817769277347482e-06,
+      "loss": 0.7731,
+      "step": 5623
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8143776185333535e-06,
+      "loss": 0.6632,
+      "step": 5624
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.810986543376633e-06,
+      "loss": 0.5543,
+      "step": 5625
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.80759605235019e-06,
+      "loss": 0.5727,
+      "step": 5626
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.804206145926807e-06,
+      "loss": 0.6398,
+      "step": 5627
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8008168245791855e-06,
+      "loss": 0.5587,
+      "step": 5628
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7974280887799395e-06,
+      "loss": 0.5553,
+      "step": 5629
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.794039939001609e-06,
+      "loss": 0.5852,
+      "step": 5630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.790652375716653e-06,
+      "loss": 0.8529,
+      "step": 5631
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.787265399397435e-06,
+      "loss": 0.5358,
+      "step": 5632
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.783879010516251e-06,
+      "loss": 0.7382,
+      "step": 5633
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.780493209545312e-06,
+      "loss": 0.5567,
+      "step": 5634
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.77710799695675e-06,
+      "loss": 0.529,
+      "step": 5635
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7737233732226005e-06,
+      "loss": 0.7446,
+      "step": 5636
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.770339338814832e-06,
+      "loss": 0.5896,
+      "step": 5637
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.766955894205326e-06,
+      "loss": 0.6594,
+      "step": 5638
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.763573039865884e-06,
+      "loss": 0.6131,
+      "step": 5639
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.760190776268217e-06,
+      "loss": 0.7142,
+      "step": 5640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.756809103883961e-06,
+      "loss": 0.8531,
+      "step": 5641
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7534280231846705e-06,
+      "loss": 0.7056,
+      "step": 5642
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.750047534641816e-06,
+      "loss": 0.8557,
+      "step": 5643
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.74666763872678e-06,
+      "loss": 0.6467,
+      "step": 5644
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.743288335910866e-06,
+      "loss": 0.6001,
+      "step": 5645
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7399096266652985e-06,
+      "loss": 0.54,
+      "step": 5646
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.736531511461217e-06,
+      "loss": 0.878,
+      "step": 5647
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.733153990769676e-06,
+      "loss": 0.6729,
+      "step": 5648
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.729777065061646e-06,
+      "loss": 0.5455,
+      "step": 5649
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7264007348080265e-06,
+      "loss": 0.8496,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.723025000479611e-06,
+      "loss": 0.6687,
+      "step": 5651
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.719649862547131e-06,
+      "loss": 0.6211,
+      "step": 5652
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.716275321481226e-06,
+      "loss": 0.5797,
+      "step": 5653
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.712901377752458e-06,
+      "loss": 0.5137,
+      "step": 5654
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.709528031831291e-06,
+      "loss": 0.5877,
+      "step": 5655
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.706155284188123e-06,
+      "loss": 0.6338,
+      "step": 5656
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.702783135293259e-06,
+      "loss": 0.7014,
+      "step": 5657
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.699411585616929e-06,
+      "loss": 0.6557,
+      "step": 5658
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.696040635629263e-06,
+      "loss": 0.5535,
+      "step": 5659
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6926702858003236e-06,
+      "loss": 0.5832,
+      "step": 5660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.689300536600082e-06,
+      "loss": 0.7087,
+      "step": 5661
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.685931388498433e-06,
+      "loss": 0.5182,
+      "step": 5662
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.682562841965174e-06,
+      "loss": 0.6084,
+      "step": 5663
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.679194897470031e-06,
+      "loss": 0.5906,
+      "step": 5664
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.67582755548264e-06,
+      "loss": 0.7074,
+      "step": 5665
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.672460816472556e-06,
+      "loss": 0.5326,
+      "step": 5666
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.669094680909248e-06,
+      "loss": 0.667,
+      "step": 5667
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.665729149262101e-06,
+      "loss": 0.9031,
+      "step": 5668
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.66236422200042e-06,
+      "loss": 0.5808,
+      "step": 5669
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.658999899593416e-06,
+      "loss": 0.5675,
+      "step": 5670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.655636182510224e-06,
+      "loss": 0.7105,
+      "step": 5671
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.652273071219892e-06,
+      "loss": 0.5511,
+      "step": 5672
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.648910566191386e-06,
+      "loss": 0.512,
+      "step": 5673
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.645548667893581e-06,
+      "loss": 0.8284,
+      "step": 5674
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.642187376795273e-06,
+      "loss": 0.6303,
+      "step": 5675
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6388266933651714e-06,
+      "loss": 0.5558,
+      "step": 5676
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6354666180719075e-06,
+      "loss": 0.6768,
+      "step": 5677
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.632107151384012e-06,
+      "loss": 0.8207,
+      "step": 5678
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.628748293769946e-06,
+      "loss": 0.6637,
+      "step": 5679
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.625390045698077e-06,
+      "loss": 0.7502,
+      "step": 5680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.622032407636697e-06,
+      "loss": 0.5739,
+      "step": 5681
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.618675380053998e-06,
+      "loss": 0.5684,
+      "step": 5682
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.615318963418095e-06,
+      "loss": 0.6018,
+      "step": 5683
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.61196315819703e-06,
+      "loss": 0.5141,
+      "step": 5684
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.608607964858737e-06,
+      "loss": 0.5666,
+      "step": 5685
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.605253383871078e-06,
+      "loss": 0.8396,
+      "step": 5686
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.601899415701828e-06,
+      "loss": 0.5657,
+      "step": 5687
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.598546060818679e-06,
+      "loss": 0.8853,
+      "step": 5688
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.595193319689228e-06,
+      "loss": 0.6973,
+      "step": 5689
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.591841192780996e-06,
+      "loss": 0.6114,
+      "step": 5690
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.588489680561413e-06,
+      "loss": 0.6197,
+      "step": 5691
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.585138783497831e-06,
+      "loss": 0.5775,
+      "step": 5692
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.581788502057502e-06,
+      "loss": 0.595,
+      "step": 5693
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.578438836707606e-06,
+      "loss": 0.7011,
+      "step": 5694
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.575089787915229e-06,
+      "loss": 0.7166,
+      "step": 5695
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5717413561473795e-06,
+      "loss": 0.6208,
+      "step": 5696
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.568393541870967e-06,
+      "loss": 0.6189,
+      "step": 5697
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5650463455528245e-06,
+      "loss": 0.8475,
+      "step": 5698
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.561699767659698e-06,
+      "loss": 0.7604,
+      "step": 5699
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.558353808658248e-06,
+      "loss": 0.576,
+      "step": 5700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.555008469015035e-06,
+      "loss": 0.8324,
+      "step": 5701
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.551663749196557e-06,
+      "loss": 0.5782,
+      "step": 5702
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.548319649669211e-06,
+      "loss": 0.541,
+      "step": 5703
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.544976170899306e-06,
+      "loss": 0.5703,
+      "step": 5704
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5416333133530675e-06,
+      "loss": 0.6627,
+      "step": 5705
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.538291077496638e-06,
+      "loss": 0.5994,
+      "step": 5706
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5349494637960734e-06,
+      "loss": 0.4556,
+      "step": 5707
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.53160847271733e-06,
+      "loss": 0.5692,
+      "step": 5708
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.528268104726293e-06,
+      "loss": 0.5652,
+      "step": 5709
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.524928360288754e-06,
+      "loss": 0.5145,
+      "step": 5710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5215892398704215e-06,
+      "loss": 0.6925,
+      "step": 5711
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.518250743936908e-06,
+      "loss": 0.6227,
+      "step": 5712
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.514912872953746e-06,
+      "loss": 0.5357,
+      "step": 5713
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.511575627386382e-06,
+      "loss": 0.9044,
+      "step": 5714
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.508239007700176e-06,
+      "loss": 0.8077,
+      "step": 5715
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.504903014360388e-06,
+      "loss": 0.585,
+      "step": 5716
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.501567647832206e-06,
+      "loss": 0.9173,
+      "step": 5717
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4982329085807255e-06,
+      "loss": 0.6014,
+      "step": 5718
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.494898797070955e-06,
+      "loss": 0.7228,
+      "step": 5719
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.491565313767805e-06,
+      "loss": 0.6754,
+      "step": 5720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.488232459136119e-06,
+      "loss": 0.703,
+      "step": 5721
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.484900233640637e-06,
+      "loss": 0.5654,
+      "step": 5722
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.481568637746022e-06,
+      "loss": 0.7997,
+      "step": 5723
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.478237671916832e-06,
+      "loss": 0.9795,
+      "step": 5724
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.474907336617554e-06,
+      "loss": 0.7645,
+      "step": 5725
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.471577632312581e-06,
+      "loss": 0.9499,
+      "step": 5726
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.468248559466221e-06,
+      "loss": 0.5485,
+      "step": 5727
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.464920118542686e-06,
+      "loss": 0.5926,
+      "step": 5728
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.461592310006108e-06,
+      "loss": 0.6274,
+      "step": 5729
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.458265134320525e-06,
+      "loss": 0.5343,
+      "step": 5730
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4549385919498975e-06,
+      "loss": 0.7348,
+      "step": 5731
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.451612683358081e-06,
+      "loss": 0.9576,
+      "step": 5732
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.448287409008854e-06,
+      "loss": 0.5144,
+      "step": 5733
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.444962769365907e-06,
+      "loss": 0.491,
+      "step": 5734
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.441638764892838e-06,
+      "loss": 0.5617,
+      "step": 5735
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.438315396053155e-06,
+      "loss": 0.5988,
+      "step": 5736
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4349926633102815e-06,
+      "loss": 0.6043,
+      "step": 5737
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.431670567127549e-06,
+      "loss": 0.5523,
+      "step": 5738
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.428349107968205e-06,
+      "loss": 0.6355,
+      "step": 5739
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.425028286295404e-06,
+      "loss": 0.7178,
+      "step": 5740
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4217081025722115e-06,
+      "loss": 0.6045,
+      "step": 5741
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.418388557261609e-06,
+      "loss": 0.7471,
+      "step": 5742
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.415069650826479e-06,
+      "loss": 0.6438,
+      "step": 5743
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.411751383729624e-06,
+      "loss": 0.6152,
+      "step": 5744
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.408433756433755e-06,
+      "loss": 0.6382,
+      "step": 5745
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.405116769401497e-06,
+      "loss": 0.5646,
+      "step": 5746
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.401800423095373e-06,
+      "loss": 0.4956,
+      "step": 5747
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.398484717977832e-06,
+      "loss": 0.6601,
+      "step": 5748
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.395169654511224e-06,
+      "loss": 0.4859,
+      "step": 5749
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.391855233157819e-06,
+      "loss": 0.6834,
+      "step": 5750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.388541454379783e-06,
+      "loss": 0.5713,
+      "step": 5751
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.385228318639204e-06,
+      "loss": 0.5111,
+      "step": 5752
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.381915826398076e-06,
+      "loss": 0.8424,
+      "step": 5753
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3786039781183105e-06,
+      "loss": 0.5075,
+      "step": 5754
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.375292774261713e-06,
+      "loss": 0.5012,
+      "step": 5755
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.371982215290014e-06,
+      "loss": 0.4866,
+      "step": 5756
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.368672301664848e-06,
+      "loss": 0.908,
+      "step": 5757
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.365363033847762e-06,
+      "loss": 0.5746,
+      "step": 5758
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.362054412300211e-06,
+      "loss": 0.5419,
+      "step": 5759
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.35874643748356e-06,
+      "loss": 0.5424,
+      "step": 5760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.355439109859088e-06,
+      "loss": 0.6066,
+      "step": 5761
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.352132429887973e-06,
+      "loss": 0.5544,
+      "step": 5762
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.348826398031313e-06,
+      "loss": 0.5146,
+      "step": 5763
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.345521014750112e-06,
+      "loss": 0.5392,
+      "step": 5764
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.34221628050529e-06,
+      "loss": 0.6119,
+      "step": 5765
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.338912195757659e-06,
+      "loss": 0.6328,
+      "step": 5766
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.335608760967961e-06,
+      "loss": 0.5488,
+      "step": 5767
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.332305976596832e-06,
+      "loss": 0.5568,
+      "step": 5768
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.329003843104831e-06,
+      "loss": 0.7749,
+      "step": 5769
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3257023609524095e-06,
+      "loss": 0.8126,
+      "step": 5770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.322401530599944e-06,
+      "loss": 0.6522,
+      "step": 5771
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.319101352507709e-06,
+      "loss": 0.5177,
+      "step": 5772
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3158018271359005e-06,
+      "loss": 0.5602,
+      "step": 5773
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.312502954944605e-06,
+      "loss": 0.5324,
+      "step": 5774
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.309204736393831e-06,
+      "loss": 0.6398,
+      "step": 5775
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.305907171943502e-06,
+      "loss": 0.6476,
+      "step": 5776
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.302610262053432e-06,
+      "loss": 0.5958,
+      "step": 5777
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.299314007183357e-06,
+      "loss": 0.5773,
+      "step": 5778
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.296018407792918e-06,
+      "loss": 0.8699,
+      "step": 5779
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2927234643416666e-06,
+      "loss": 0.6293,
+      "step": 5780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.289429177289057e-06,
+      "loss": 0.8907,
+      "step": 5781
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2861355470944564e-06,
+      "loss": 0.6487,
+      "step": 5782
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.28284257421714e-06,
+      "loss": 0.5294,
+      "step": 5783
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.279550259116297e-06,
+      "loss": 0.563,
+      "step": 5784
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.27625860225101e-06,
+      "loss": 0.5111,
+      "step": 5785
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.272967604080283e-06,
+      "loss": 0.6222,
+      "step": 5786
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.269677265063022e-06,
+      "loss": 0.562,
+      "step": 5787
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.26638758565805e-06,
+      "loss": 0.6116,
+      "step": 5788
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2630985663240805e-06,
+      "loss": 0.6189,
+      "step": 5789
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.259810207519753e-06,
+      "loss": 0.589,
+      "step": 5790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.256522509703605e-06,
+      "loss": 0.5808,
+      "step": 5791
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.253235473334087e-06,
+      "loss": 0.7147,
+      "step": 5792
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.249949098869546e-06,
+      "loss": 0.6716,
+      "step": 5793
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.246663386768254e-06,
+      "loss": 0.628,
+      "step": 5794
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.243378337488383e-06,
+      "loss": 0.7451,
+      "step": 5795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.240093951488006e-06,
+      "loss": 0.9336,
+      "step": 5796
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.236810229225108e-06,
+      "loss": 0.6018,
+      "step": 5797
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.233527171157587e-06,
+      "loss": 0.6131,
+      "step": 5798
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.230244777743243e-06,
+      "loss": 0.7838,
+      "step": 5799
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.226963049439781e-06,
+      "loss": 0.535,
+      "step": 5800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.223681986704818e-06,
+      "loss": 0.7625,
+      "step": 5801
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.220401589995876e-06,
+      "loss": 0.6357,
+      "step": 5802
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.21712185977039e-06,
+      "loss": 0.8146,
+      "step": 5803
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2138427964856885e-06,
+      "loss": 0.4771,
+      "step": 5804
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2105644005990185e-06,
+      "loss": 0.5832,
+      "step": 5805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.207286672567531e-06,
+      "loss": 0.8117,
+      "step": 5806
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.204009612848288e-06,
+      "loss": 0.5776,
+      "step": 5807
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.200733221898244e-06,
+      "loss": 0.8716,
+      "step": 5808
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.197457500174276e-06,
+      "loss": 0.6443,
+      "step": 5809
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.194182448133163e-06,
+      "loss": 0.5135,
+      "step": 5810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.190908066231591e-06,
+      "loss": 0.4861,
+      "step": 5811
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.187634354926139e-06,
+      "loss": 0.6648,
+      "step": 5812
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.184361314673318e-06,
+      "loss": 0.6366,
+      "step": 5813
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.181088945929531e-06,
+      "loss": 0.7325,
+      "step": 5814
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.177817249151079e-06,
+      "loss": 0.7193,
+      "step": 5815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.174546224794183e-06,
+      "loss": 0.5845,
+      "step": 5816
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.171275873314969e-06,
+      "loss": 0.6273,
+      "step": 5817
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.168006195169466e-06,
+      "loss": 0.5345,
+      "step": 5818
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1647371908136026e-06,
+      "loss": 0.6783,
+      "step": 5819
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1614688607032234e-06,
+      "loss": 0.8524,
+      "step": 5820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1582012052940755e-06,
+      "loss": 0.5866,
+      "step": 5821
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.154934225041817e-06,
+      "loss": 0.5173,
+      "step": 5822
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.151667920401998e-06,
+      "loss": 0.4918,
+      "step": 5823
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.148402291830086e-06,
+      "loss": 0.7758,
+      "step": 5824
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.145137339781453e-06,
+      "loss": 0.8228,
+      "step": 5825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.141873064711378e-06,
+      "loss": 0.4851,
+      "step": 5826
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.138609467075035e-06,
+      "loss": 0.5803,
+      "step": 5827
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.135346547327516e-06,
+      "loss": 0.7165,
+      "step": 5828
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.132084305923813e-06,
+      "loss": 0.5088,
+      "step": 5829
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.128822743318822e-06,
+      "loss": 0.8669,
+      "step": 5830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.125561859967349e-06,
+      "loss": 0.7883,
+      "step": 5831
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.122301656324102e-06,
+      "loss": 0.6632,
+      "step": 5832
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.119042132843694e-06,
+      "loss": 0.6687,
+      "step": 5833
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.115783289980649e-06,
+      "loss": 0.5035,
+      "step": 5834
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1125251281893824e-06,
+      "loss": 0.7611,
+      "step": 5835
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.109267647924229e-06,
+      "loss": 0.5741,
+      "step": 5836
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.106010849639422e-06,
+      "loss": 0.5865,
+      "step": 5837
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.102754733789104e-06,
+      "loss": 0.6757,
+      "step": 5838
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0994993008273086e-06,
+      "loss": 0.7603,
+      "step": 5839
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.096244551207994e-06,
+      "loss": 0.7121,
+      "step": 5840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.092990485385009e-06,
+      "loss": 0.7157,
+      "step": 5841
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.089737103812116e-06,
+      "loss": 0.91,
+      "step": 5842
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.086484406942973e-06,
+      "loss": 0.8511,
+      "step": 5843
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.083232395231147e-06,
+      "loss": 0.764,
+      "step": 5844
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.079981069130111e-06,
+      "loss": 0.5707,
+      "step": 5845
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.076730429093246e-06,
+      "loss": 0.641,
+      "step": 5846
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.073480475573824e-06,
+      "loss": 0.5283,
+      "step": 5847
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0702312090250326e-06,
+      "loss": 0.8527,
+      "step": 5848
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.066982629899961e-06,
+      "loss": 0.5715,
+      "step": 5849
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.063734738651602e-06,
+      "loss": 0.6066,
+      "step": 5850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.060487535732854e-06,
+      "loss": 0.6689,
+      "step": 5851
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.057241021596515e-06,
+      "loss": 0.5956,
+      "step": 5852
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.053995196695297e-06,
+      "loss": 0.5953,
+      "step": 5853
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.050750061481799e-06,
+      "loss": 0.5609,
+      "step": 5854
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.047505616408539e-06,
+      "loss": 0.6234,
+      "step": 5855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.044261861927933e-06,
+      "loss": 0.581,
+      "step": 5856
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.041018798492304e-06,
+      "loss": 0.6802,
+      "step": 5857
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.037776426553869e-06,
+      "loss": 0.8087,
+      "step": 5858
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.034534746564759e-06,
+      "loss": 0.5351,
+      "step": 5859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.031293758977005e-06,
+      "loss": 0.5256,
+      "step": 5860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.028053464242546e-06,
+      "loss": 0.9994,
+      "step": 5861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.02481386281321e-06,
+      "loss": 0.6867,
+      "step": 5862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.021574955140744e-06,
+      "loss": 0.7596,
+      "step": 5863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.018336741676791e-06,
+      "loss": 0.6386,
+      "step": 5864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0150992228729034e-06,
+      "loss": 0.5203,
+      "step": 5865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.011862399180523e-06,
+      "loss": 0.6048,
+      "step": 5866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.008626271051002e-06,
+      "loss": 0.51,
+      "step": 5867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0053908389356135e-06,
+      "loss": 0.5015,
+      "step": 5868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.002156103285501e-06,
+      "loss": 0.5343,
+      "step": 5869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.998922064551732e-06,
+      "loss": 0.7271,
+      "step": 5870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.995688723185273e-06,
+      "loss": 0.4994,
+      "step": 5871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.992456079636997e-06,
+      "loss": 0.8005,
+      "step": 5872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.989224134357664e-06,
+      "loss": 0.4976,
+      "step": 5873
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.985992887797953e-06,
+      "loss": 0.7344,
+      "step": 5874
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9827623404084394e-06,
+      "loss": 0.8445,
+      "step": 5875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.979532492639606e-06,
+      "loss": 0.5739,
+      "step": 5876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9763033449418265e-06,
+      "loss": 0.6495,
+      "step": 5877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.973074897765387e-06,
+      "loss": 0.8117,
+      "step": 5878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.969847151560475e-06,
+      "loss": 0.7405,
+      "step": 5879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9666201067771806e-06,
+      "loss": 0.5599,
+      "step": 5880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.963393763865485e-06,
+      "loss": 0.905,
+      "step": 5881
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9601681232752875e-06,
+      "loss": 0.5661,
+      "step": 5882
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.95694318545638e-06,
+      "loss": 0.5985,
+      "step": 5883
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.953718950858464e-06,
+      "loss": 0.8765,
+      "step": 5884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.950495419931128e-06,
+      "loss": 0.5253,
+      "step": 5885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.947272593123875e-06,
+      "loss": 0.9429,
+      "step": 5886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.944050470886117e-06,
+      "loss": 0.643,
+      "step": 5887
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.940829053667146e-06,
+      "loss": 0.4926,
+      "step": 5888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.93760834191617e-06,
+      "loss": 0.6436,
+      "step": 5889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.934388336082298e-06,
+      "loss": 0.5475,
+      "step": 5890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.931169036614543e-06,
+      "loss": 0.6651,
+      "step": 5891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.927950443961805e-06,
+      "loss": 0.5691,
+      "step": 5892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.924732558572901e-06,
+      "loss": 0.6699,
+      "step": 5893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.921515380896543e-06,
+      "loss": 0.5524,
+      "step": 5894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9182989113813494e-06,
+      "loss": 0.7051,
+      "step": 5895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.915083150475828e-06,
+      "loss": 0.5235,
+      "step": 5896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9118680986284e-06,
+      "loss": 0.8856,
+      "step": 5897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.908653756287382e-06,
+      "loss": 0.5507,
+      "step": 5898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.905440123900996e-06,
+      "loss": 0.7869,
+      "step": 5899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9022272019173555e-06,
+      "loss": 0.5171,
+      "step": 5900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.899014990784485e-06,
+      "loss": 0.5757,
+      "step": 5901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.895803490950307e-06,
+      "loss": 0.5919,
+      "step": 5902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.892592702862645e-06,
+      "loss": 0.8541,
+      "step": 5903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8893826269692145e-06,
+      "loss": 0.5441,
+      "step": 5904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.886173263717647e-06,
+      "loss": 0.8086,
+      "step": 5905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.88296461355547e-06,
+      "loss": 0.5366,
+      "step": 5906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.879756676930098e-06,
+      "loss": 0.5848,
+      "step": 5907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.876549454288864e-06,
+      "loss": 0.6161,
+      "step": 5908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.873342946078993e-06,
+      "loss": 0.4179,
+      "step": 5909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.870137152747613e-06,
+      "loss": 0.5063,
+      "step": 5910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.866932074741745e-06,
+      "loss": 0.5151,
+      "step": 5911
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.863727712508318e-06,
+      "loss": 0.6124,
+      "step": 5912
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8605240664941626e-06,
+      "loss": 0.6126,
+      "step": 5913
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8573211371460064e-06,
+      "loss": 0.6402,
+      "step": 5914
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.85411892491047e-06,
+      "loss": 0.5286,
+      "step": 5915
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.850917430234086e-06,
+      "loss": 0.5678,
+      "step": 5916
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8477166535632815e-06,
+      "loss": 0.6194,
+      "step": 5917
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.844516595344386e-06,
+      "loss": 0.4943,
+      "step": 5918
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.84131725602362e-06,
+      "loss": 0.6208,
+      "step": 5919
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.838118636047113e-06,
+      "loss": 0.618,
+      "step": 5920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8349207358608925e-06,
+      "loss": 0.5969,
+      "step": 5921
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.831723555910885e-06,
+      "loss": 0.5948,
+      "step": 5922
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.828527096642915e-06,
+      "loss": 0.7367,
+      "step": 5923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.825331358502707e-06,
+      "loss": 0.6134,
+      "step": 5924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.822136341935891e-06,
+      "loss": 0.6214,
+      "step": 5925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.818942047387983e-06,
+      "loss": 0.5321,
+      "step": 5926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.81574847530441e-06,
+      "loss": 0.7277,
+      "step": 5927
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.812555626130495e-06,
+      "loss": 0.5411,
+      "step": 5928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.809363500311463e-06,
+      "loss": 0.6424,
+      "step": 5929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8061720982924255e-06,
+      "loss": 0.6339,
+      "step": 5930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.802981420518409e-06,
+      "loss": 0.9069,
+      "step": 5931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.799791467434331e-06,
+      "loss": 0.6135,
+      "step": 5932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7966022394850154e-06,
+      "loss": 0.536,
+      "step": 5933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7934137371151676e-06,
+      "loss": 0.5608,
+      "step": 5934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7902259607694105e-06,
+      "loss": 0.6259,
+      "step": 5935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.787038910892257e-06,
+      "loss": 0.5414,
+      "step": 5936
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.783852587928123e-06,
+      "loss": 0.6267,
+      "step": 5937
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.780666992321315e-06,
+      "loss": 0.5894,
+      "step": 5938
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.777482124516044e-06,
+      "loss": 0.6553,
+      "step": 5939
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.774297984956422e-06,
+      "loss": 0.612,
+      "step": 5940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.771114574086454e-06,
+      "loss": 0.5502,
+      "step": 5941
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.767931892350046e-06,
+      "loss": 0.5095,
+      "step": 5942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.764749940191002e-06,
+      "loss": 0.5957,
+      "step": 5943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.761568718053026e-06,
+      "loss": 0.8321,
+      "step": 5944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.75838822637972e-06,
+      "loss": 0.5753,
+      "step": 5945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.755208465614576e-06,
+      "loss": 0.6457,
+      "step": 5946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.752029436200994e-06,
+      "loss": 0.6738,
+      "step": 5947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.748851138582269e-06,
+      "loss": 0.8121,
+      "step": 5948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.745673573201597e-06,
+      "loss": 0.5841,
+      "step": 5949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.742496740502061e-06,
+      "loss": 0.6802,
+      "step": 5950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.739320640926655e-06,
+      "loss": 0.6159,
+      "step": 5951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.736145274918262e-06,
+      "loss": 0.4776,
+      "step": 5952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7329706429196705e-06,
+      "loss": 0.6303,
+      "step": 5953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.729796745373557e-06,
+      "loss": 0.5495,
+      "step": 5954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7266235827225004e-06,
+      "loss": 0.6298,
+      "step": 5955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.723451155408981e-06,
+      "loss": 0.6142,
+      "step": 5956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.720279463875374e-06,
+      "loss": 0.5917,
+      "step": 5957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7171085085639444e-06,
+      "loss": 0.6336,
+      "step": 5958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.713938289916861e-06,
+      "loss": 0.5422,
+      "step": 5959
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.710768808376201e-06,
+      "loss": 0.6908,
+      "step": 5960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.707600064383915e-06,
+      "loss": 0.6467,
+      "step": 5961
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.704432058381869e-06,
+      "loss": 0.5494,
+      "step": 5962
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.701264790811821e-06,
+      "loss": 0.5763,
+      "step": 5963
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.698098262115428e-06,
+      "loss": 0.8993,
+      "step": 5964
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.694932472734234e-06,
+      "loss": 0.5463,
+      "step": 5965
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.691767423109692e-06,
+      "loss": 0.8359,
+      "step": 5966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6886031136831465e-06,
+      "loss": 0.5829,
+      "step": 5967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.685439544895844e-06,
+      "loss": 0.6203,
+      "step": 5968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.682276717188915e-06,
+      "loss": 0.4668,
+      "step": 5969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6791146310034e-06,
+      "loss": 0.5444,
+      "step": 5970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.675953286780229e-06,
+      "loss": 0.7213,
+      "step": 5971
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.672792684960237e-06,
+      "loss": 0.4908,
+      "step": 5972
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.66963282598414e-06,
+      "loss": 0.5497,
+      "step": 5973
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.666473710292562e-06,
+      "loss": 0.6355,
+      "step": 5974
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.663315338326022e-06,
+      "loss": 0.643,
+      "step": 5975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.660157710524939e-06,
+      "loss": 0.812,
+      "step": 5976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6570008273296154e-06,
+      "loss": 0.5844,
+      "step": 5977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.653844689180255e-06,
+      "loss": 0.6254,
+      "step": 5978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.650689296516977e-06,
+      "loss": 0.5986,
+      "step": 5979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.647534649779764e-06,
+      "loss": 0.5972,
+      "step": 5980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.644380749408516e-06,
+      "loss": 0.793,
+      "step": 5981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.641227595843024e-06,
+      "loss": 0.779,
+      "step": 5982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.638075189522978e-06,
+      "loss": 0.5451,
+      "step": 5983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.634923530887952e-06,
+      "loss": 0.6035,
+      "step": 5984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.631772620377431e-06,
+      "loss": 0.9033,
+      "step": 5985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.628622458430784e-06,
+      "loss": 0.8368,
+      "step": 5986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.625473045487286e-06,
+      "loss": 0.7952,
+      "step": 5987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.622324381986094e-06,
+      "loss": 0.5461,
+      "step": 5988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.619176468366274e-06,
+      "loss": 0.5493,
+      "step": 5989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.61602930506678e-06,
+      "loss": 0.6551,
+      "step": 5990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.612882892526467e-06,
+      "loss": 0.6638,
+      "step": 5991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.609737231184075e-06,
+      "loss": 0.5759,
+      "step": 5992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.606592321478248e-06,
+      "loss": 0.6165,
+      "step": 5993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6034481638475235e-06,
+      "loss": 0.5899,
+      "step": 5994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6003047587303376e-06,
+      "loss": 0.5825,
+      "step": 5995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.597162106565005e-06,
+      "loss": 0.6662,
+      "step": 5996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.594020207789761e-06,
+      "loss": 0.5284,
+      "step": 5997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.590879062842722e-06,
+      "loss": 0.6281,
+      "step": 5998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.587738672161891e-06,
+      "loss": 0.5561,
+      "step": 5999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.58459903618518e-06,
+      "loss": 0.7374,
+      "step": 6000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.58146015535039e-06,
+      "loss": 0.9007,
+      "step": 6001
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.578322030095221e-06,
+      "loss": 0.5046,
+      "step": 6002
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.575184660857257e-06,
+      "loss": 0.6142,
+      "step": 6003
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.572048048073986e-06,
+      "loss": 0.611,
+      "step": 6004
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5689121921827895e-06,
+      "loss": 0.5422,
+      "step": 6005
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.565777093620943e-06,
+      "loss": 0.7373,
+      "step": 6006
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.562642752825611e-06,
+      "loss": 0.6456,
+      "step": 6007
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5595091702338575e-06,
+      "loss": 0.6549,
+      "step": 6008
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.556376346282643e-06,
+      "loss": 0.7037,
+      "step": 6009
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5532442814088195e-06,
+      "loss": 0.8389,
+      "step": 6010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5501129760491266e-06,
+      "loss": 0.4799,
+      "step": 6011
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.54698243064021e-06,
+      "loss": 0.5687,
+      "step": 6012
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.543852645618601e-06,
+      "loss": 0.6542,
+      "step": 6013
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5407236214207285e-06,
+      "loss": 0.5184,
+      "step": 6014
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.537595358482914e-06,
+      "loss": 0.5418,
+      "step": 6015
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.534467857241372e-06,
+      "loss": 0.5019,
+      "step": 6016
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.531341118132219e-06,
+      "loss": 0.7606,
+      "step": 6017
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5282151415914466e-06,
+      "loss": 0.6993,
+      "step": 6018
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.525089928054958e-06,
+      "loss": 0.5018,
+      "step": 6019
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.521965477958542e-06,
+      "loss": 0.5626,
+      "step": 6020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.518841791737888e-06,
+      "loss": 0.6725,
+      "step": 6021
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.515718869828565e-06,
+      "loss": 0.89,
+      "step": 6022
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.512596712666047e-06,
+      "loss": 0.6007,
+      "step": 6023
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5094753206857e-06,
+      "loss": 0.6024,
+      "step": 6024
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.506354694322783e-06,
+      "loss": 0.9837,
+      "step": 6025
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.503234834012441e-06,
+      "loss": 0.5537,
+      "step": 6026
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.500115740189721e-06,
+      "loss": 0.7591,
+      "step": 6027
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.49699741328956e-06,
+      "loss": 0.6643,
+      "step": 6028
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.493879853746792e-06,
+      "loss": 0.6134,
+      "step": 6029
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.490763061996133e-06,
+      "loss": 0.5855,
+      "step": 6030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.487647038472201e-06,
+      "loss": 0.5375,
+      "step": 6031
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.484531783609507e-06,
+      "loss": 0.8366,
+      "step": 6032
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.481417297842452e-06,
+      "loss": 0.5615,
+      "step": 6033
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.47830358160533e-06,
+      "loss": 0.5059,
+      "step": 6034
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.475190635332329e-06,
+      "loss": 0.5934,
+      "step": 6035
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.472078459457533e-06,
+      "loss": 0.5729,
+      "step": 6036
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.468967054414905e-06,
+      "loss": 0.6418,
+      "step": 6037
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4658564206383145e-06,
+      "loss": 0.8084,
+      "step": 6038
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.46274655856152e-06,
+      "loss": 0.6197,
+      "step": 6039
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4596374686181735e-06,
+      "loss": 0.805,
+      "step": 6040
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.456529151241809e-06,
+      "loss": 0.5458,
+      "step": 6041
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.453421606865869e-06,
+      "loss": 0.5161,
+      "step": 6042
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4503148359236734e-06,
+      "loss": 0.5965,
+      "step": 6043
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4472088388484505e-06,
+      "loss": 0.6192,
+      "step": 6044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4441036160733e-06,
+      "loss": 0.8597,
+      "step": 6045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.440999168031229e-06,
+      "loss": 0.7095,
+      "step": 6046
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.437895495155133e-06,
+      "loss": 0.4664,
+      "step": 6047
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.434792597877803e-06,
+      "loss": 0.5481,
+      "step": 6048
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4316904766319095e-06,
+      "loss": 0.5335,
+      "step": 6049
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.428589131850025e-06,
+      "loss": 0.5256,
+      "step": 6050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.425488563964615e-06,
+      "loss": 0.6144,
+      "step": 6051
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.422388773408029e-06,
+      "loss": 0.5687,
+      "step": 6052
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.419289760612517e-06,
+      "loss": 0.5507,
+      "step": 6053
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.416191526010213e-06,
+      "loss": 0.5444,
+      "step": 6054
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4130940700331444e-06,
+      "loss": 0.5996,
+      "step": 6055
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.409997393113237e-06,
+      "loss": 0.7811,
+      "step": 6056
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.406901495682293e-06,
+      "loss": 0.5413,
+      "step": 6057
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.403806378172021e-06,
+      "loss": 0.807,
+      "step": 6058
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.400712041014012e-06,
+      "loss": 0.5582,
+      "step": 6059
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.397618484639754e-06,
+      "loss": 0.5366,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.394525709480618e-06,
+      "loss": 0.8377,
+      "step": 6061
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.391433715967875e-06,
+      "loss": 0.6357,
+      "step": 6062
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.388342504532681e-06,
+      "loss": 0.4878,
+      "step": 6063
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.38525207560609e-06,
+      "loss": 0.5748,
+      "step": 6064
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.382162429619035e-06,
+      "loss": 0.5273,
+      "step": 6065
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3790735670023485e-06,
+      "loss": 0.6482,
+      "step": 6066
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.375985488186755e-06,
+      "loss": 0.5058,
+      "step": 6067
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.372898193602867e-06,
+      "loss": 0.6123,
+      "step": 6068
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.369811683681182e-06,
+      "loss": 0.611,
+      "step": 6069
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.366725958852093e-06,
+      "loss": 0.5439,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.363641019545895e-06,
+      "loss": 0.6692,
+      "step": 6071
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.360556866192751e-06,
+      "loss": 0.6757,
+      "step": 6072
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.357473499222731e-06,
+      "loss": 0.851,
+      "step": 6073
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.354390919065788e-06,
+      "loss": 0.5815,
+      "step": 6074
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.351309126151773e-06,
+      "loss": 0.5817,
+      "step": 6075
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3482281209104135e-06,
+      "loss": 0.4985,
+      "step": 6076
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.345147903771339e-06,
+      "loss": 0.5743,
+      "step": 6077
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.342068475164065e-06,
+      "loss": 0.579,
+      "step": 6078
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.338989835518003e-06,
+      "loss": 0.5735,
+      "step": 6079
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.335911985262439e-06,
+      "loss": 0.5606,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.332834924826565e-06,
+      "loss": 0.6829,
+      "step": 6081
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.329758654639455e-06,
+      "loss": 0.7416,
+      "step": 6082
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.326683175130077e-06,
+      "loss": 0.7401,
+      "step": 6083
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.323608486727283e-06,
+      "loss": 0.6562,
+      "step": 6084
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.32053458985982e-06,
+      "loss": 0.5163,
+      "step": 6085
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.317461484956322e-06,
+      "loss": 0.672,
+      "step": 6086
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.314389172445317e-06,
+      "loss": 0.5651,
+      "step": 6087
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.311317652755207e-06,
+      "loss": 0.4913,
+      "step": 6088
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.308246926314307e-06,
+      "loss": 0.7042,
+      "step": 6089
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.30517699355081e-06,
+      "loss": 0.6401,
+      "step": 6090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3021078548927894e-06,
+      "loss": 0.6182,
+      "step": 6091
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299039510768222e-06,
+      "loss": 0.5863,
+      "step": 6092
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.295971961604965e-06,
+      "loss": 0.4996,
+      "step": 6093
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.292905207830773e-06,
+      "loss": 0.576,
+      "step": 6094
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.289839249873279e-06,
+      "loss": 0.9288,
+      "step": 6095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2867740881600115e-06,
+      "loss": 0.5864,
+      "step": 6096
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.28370972311839e-06,
+      "loss": 0.8423,
+      "step": 6097
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.28064615517572e-06,
+      "loss": 0.5321,
+      "step": 6098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.277583384759192e-06,
+      "loss": 0.588,
+      "step": 6099
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.274521412295891e-06,
+      "loss": 0.486,
+      "step": 6100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.271460238212788e-06,
+      "loss": 0.5371,
+      "step": 6101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.268399862936749e-06,
+      "loss": 0.6033,
+      "step": 6102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.265340286894516e-06,
+      "loss": 0.5959,
+      "step": 6103
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2622815105127276e-06,
+      "loss": 0.6145,
+      "step": 6104
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2592235342179135e-06,
+      "loss": 0.5975,
+      "step": 6105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.256166358436486e-06,
+      "loss": 0.505,
+      "step": 6106
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.253109983594749e-06,
+      "loss": 0.5933,
+      "step": 6107
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2500544101188935e-06,
+      "loss": 0.5277,
+      "step": 6108
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.246999638435002e-06,
+      "loss": 0.4813,
+      "step": 6109
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.243945668969036e-06,
+      "loss": 0.7552,
+      "step": 6110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.240892502146855e-06,
+      "loss": 0.6144,
+      "step": 6111
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.237840138394202e-06,
+      "loss": 0.6586,
+      "step": 6112
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.234788578136714e-06,
+      "loss": 0.8425,
+      "step": 6113
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.231737821799903e-06,
+      "loss": 0.5257,
+      "step": 6114
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2286878698091805e-06,
+      "loss": 0.6059,
+      "step": 6115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.225638722589841e-06,
+      "loss": 0.6624,
+      "step": 6116
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2225903805670744e-06,
+      "loss": 0.6043,
+      "step": 6117
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.219542844165941e-06,
+      "loss": 0.5248,
+      "step": 6118
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216496113811407e-06,
+      "loss": 0.5703,
+      "step": 6119
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213450189928315e-06,
+      "loss": 0.5894,
+      "step": 6120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210405072941406e-06,
+      "loss": 0.723,
+      "step": 6121
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207360763275292e-06,
+      "loss": 0.7224,
+      "step": 6122
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.204317261354488e-06,
+      "loss": 0.6735,
+      "step": 6123
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.201274567603387e-06,
+      "loss": 0.6279,
+      "step": 6124
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.198232682446275e-06,
+      "loss": 0.5123,
+      "step": 6125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.195191606307321e-06,
+      "loss": 0.6581,
+      "step": 6126
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1921513396105854e-06,
+      "loss": 0.5767,
+      "step": 6127
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.189111882780014e-06,
+      "loss": 0.9516,
+      "step": 6128
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.186073236239434e-06,
+      "loss": 0.9004,
+      "step": 6129
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.183035400412567e-06,
+      "loss": 0.6148,
+      "step": 6130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.17999837572302e-06,
+      "loss": 0.543,
+      "step": 6131
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.176962162594289e-06,
+      "loss": 0.902,
+      "step": 6132
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.173926761449747e-06,
+      "loss": 0.6539,
+      "step": 6133
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.170892172712665e-06,
+      "loss": 0.5251,
+      "step": 6134
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.167858396806195e-06,
+      "loss": 0.7705,
+      "step": 6135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.164825434153381e-06,
+      "loss": 0.5704,
+      "step": 6136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.161793285177142e-06,
+      "loss": 0.5832,
+      "step": 6137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.158761950300297e-06,
+      "loss": 0.5757,
+      "step": 6138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.155731429945543e-06,
+      "loss": 0.5079,
+      "step": 6139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.152701724535471e-06,
+      "loss": 0.7525,
+      "step": 6140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.149672834492546e-06,
+      "loss": 0.6802,
+      "step": 6141
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.14664476023913e-06,
+      "loss": 0.7543,
+      "step": 6142
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1436175021974686e-06,
+      "loss": 0.8655,
+      "step": 6143
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.140591060789693e-06,
+      "loss": 0.69,
+      "step": 6144
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.137565436437818e-06,
+      "loss": 0.5825,
+      "step": 6145
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.134540629563751e-06,
+      "loss": 0.6007,
+      "step": 6146
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.131516640589283e-06,
+      "loss": 0.5045,
+      "step": 6147
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.12849346993608e-06,
+      "loss": 0.7312,
+      "step": 6148
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.125471118025711e-06,
+      "loss": 0.6676,
+      "step": 6149
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1224495852796194e-06,
+      "loss": 0.5886,
+      "step": 6150
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.119428872119144e-06,
+      "loss": 0.5738,
+      "step": 6151
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.116408978965495e-06,
+      "loss": 0.5415,
+      "step": 6152
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.11338990623978e-06,
+      "loss": 0.5413,
+      "step": 6153
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.110371654362989e-06,
+      "loss": 0.5168,
+      "step": 6154
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.107354223756002e-06,
+      "loss": 0.5737,
+      "step": 6155
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.104337614839571e-06,
+      "loss": 0.6629,
+      "step": 6156
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1013218280343455e-06,
+      "loss": 0.7745,
+      "step": 6157
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0983068637608594e-06,
+      "loss": 0.6595,
+      "step": 6158
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.095292722439531e-06,
+      "loss": 0.5183,
+      "step": 6159
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0922794044906566e-06,
+      "loss": 0.603,
+      "step": 6160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.089266910334425e-06,
+      "loss": 0.8681,
+      "step": 6161
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.086255240390912e-06,
+      "loss": 0.6007,
+      "step": 6162
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.083244395080071e-06,
+      "loss": 0.5243,
+      "step": 6163
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.080234374821749e-06,
+      "loss": 0.5117,
+      "step": 6164
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.07722518003567e-06,
+      "loss": 0.8592,
+      "step": 6165
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.07421681114145e-06,
+      "loss": 0.6698,
+      "step": 6166
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.071209268558581e-06,
+      "loss": 0.7276,
+      "step": 6167
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0682025527064486e-06,
+      "loss": 0.5258,
+      "step": 6168
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0651966640043176e-06,
+      "loss": 0.6794,
+      "step": 6169
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.062191602871344e-06,
+      "loss": 0.6156,
+      "step": 6170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.059187369726556e-06,
+      "loss": 0.6585,
+      "step": 6171
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.056183964988879e-06,
+      "loss": 0.5696,
+      "step": 6172
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.053181389077115e-06,
+      "loss": 0.587,
+      "step": 6173
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.050179642409962e-06,
+      "loss": 0.6656,
+      "step": 6174
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.047178725405981e-06,
+      "loss": 0.5978,
+      "step": 6175
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.044178638483638e-06,
+      "loss": 0.7827,
+      "step": 6176
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.041179382061272e-06,
+      "loss": 0.5253,
+      "step": 6177
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.038180956557116e-06,
+      "loss": 0.77,
+      "step": 6178
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.035183362389272e-06,
+      "loss": 0.5558,
+      "step": 6179
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.032186599975735e-06,
+      "loss": 0.5939,
+      "step": 6180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.02919066973439e-06,
+      "loss": 0.8358,
+      "step": 6181
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.026195572083e-06,
+      "loss": 0.5116,
+      "step": 6182
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0232013074392065e-06,
+      "loss": 0.8537,
+      "step": 6183
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.020207876220541e-06,
+      "loss": 0.5525,
+      "step": 6184
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.01721527884442e-06,
+      "loss": 0.8766,
+      "step": 6185
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.014223515728142e-06,
+      "loss": 0.5656,
+      "step": 6186
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0112325872888835e-06,
+      "loss": 0.7982,
+      "step": 6187
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.008242493943714e-06,
+      "loss": 0.8508,
+      "step": 6188
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.00525323610958e-06,
+      "loss": 0.5731,
+      "step": 6189
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0022648142033195e-06,
+      "loss": 0.6114,
+      "step": 6190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.999277228641639e-06,
+      "loss": 0.5821,
+      "step": 6191
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.996290479841143e-06,
+      "loss": 0.8501,
+      "step": 6192
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9933045682183134e-06,
+      "loss": 0.5189,
+      "step": 6193
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.99031949418952e-06,
+      "loss": 0.6673,
+      "step": 6194
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.987335258171002e-06,
+      "loss": 0.8983,
+      "step": 6195
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.984351860578898e-06,
+      "loss": 0.596,
+      "step": 6196
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.981369301829222e-06,
+      "loss": 0.4833,
+      "step": 6197
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.978387582337871e-06,
+      "loss": 0.869,
+      "step": 6198
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.975406702520628e-06,
+      "loss": 0.6009,
+      "step": 6199
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.972426662793156e-06,
+      "loss": 0.8076,
+      "step": 6200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.969447463571006e-06,
+      "loss": 0.7239,
+      "step": 6201
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.966469105269599e-06,
+      "loss": 0.7514,
+      "step": 6202
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.963491588304252e-06,
+      "loss": 0.5658,
+      "step": 6203
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.960514913090162e-06,
+      "loss": 0.7763,
+      "step": 6204
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.957539080042408e-06,
+      "loss": 0.8775,
+      "step": 6205
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.954564089575942e-06,
+      "loss": 0.4278,
+      "step": 6206
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.951589942105613e-06,
+      "loss": 0.7292,
+      "step": 6207
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.948616638046144e-06,
+      "loss": 0.5372,
+      "step": 6208
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.945644177812149e-06,
+      "loss": 0.6833,
+      "step": 6209
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.942672561818108e-06,
+      "loss": 0.7083,
+      "step": 6210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9397017904784e-06,
+      "loss": 0.6708,
+      "step": 6211
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.936731864207275e-06,
+      "loss": 0.8064,
+      "step": 6212
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.933762783418878e-06,
+      "loss": 0.8824,
+      "step": 6213
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.930794548527216e-06,
+      "loss": 0.5943,
+      "step": 6214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.927827159946196e-06,
+      "loss": 0.6022,
+      "step": 6215
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.924860618089601e-06,
+      "loss": 0.6522,
+      "step": 6216
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9218949233710944e-06,
+      "loss": 0.5564,
+      "step": 6217
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.918930076204224e-06,
+      "loss": 0.6944,
+      "step": 6218
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.915966077002417e-06,
+      "loss": 0.5068,
+      "step": 6219
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9130029261789885e-06,
+      "loss": 0.616,
+      "step": 6220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.910040624147121e-06,
+      "loss": 0.6148,
+      "step": 6221
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.907079171319894e-06,
+      "loss": 0.5186,
+      "step": 6222
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.904118568110261e-06,
+      "loss": 0.542,
+      "step": 6223
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.901158814931062e-06,
+      "loss": 0.5123,
+      "step": 6224
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.898199912195009e-06,
+      "loss": 0.8383,
+      "step": 6225
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8952418603147035e-06,
+      "loss": 0.56,
+      "step": 6226
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8922846597026286e-06,
+      "loss": 0.6322,
+      "step": 6227
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.889328310771148e-06,
+      "loss": 0.5601,
+      "step": 6228
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.886372813932499e-06,
+      "loss": 0.4858,
+      "step": 6229
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.883418169598808e-06,
+      "loss": 0.5283,
+      "step": 6230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.880464378182083e-06,
+      "loss": 0.8338,
+      "step": 6231
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8775114400942135e-06,
+      "loss": 0.5677,
+      "step": 6232
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.874559355746959e-06,
+      "loss": 0.7213,
+      "step": 6233
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.871608125551973e-06,
+      "loss": 0.5728,
+      "step": 6234
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.868657749920785e-06,
+      "loss": 0.7626,
+      "step": 6235
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865708229264803e-06,
+      "loss": 0.5259,
+      "step": 6236
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.862759563995322e-06,
+      "loss": 0.6928,
+      "step": 6237
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.859811754523511e-06,
+      "loss": 0.5245,
+      "step": 6238
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.856864801260428e-06,
+      "loss": 0.7128,
+      "step": 6239
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.853918704616999e-06,
+      "loss": 0.519,
+      "step": 6240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.85097346500404e-06,
+      "loss": 0.7774,
+      "step": 6241
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.848029082832245e-06,
+      "loss": 0.6306,
+      "step": 6242
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.845085558512193e-06,
+      "loss": 0.5889,
+      "step": 6243
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.842142892454333e-06,
+      "loss": 0.7742,
+      "step": 6244
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.839201085069002e-06,
+      "loss": 0.5318,
+      "step": 6245
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.836260136766418e-06,
+      "loss": 0.8073,
+      "step": 6246
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.833320047956677e-06,
+      "loss": 0.712,
+      "step": 6247
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.83038081904975e-06,
+      "loss": 0.5278,
+      "step": 6248
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8274424504554975e-06,
+      "loss": 0.6301,
+      "step": 6249
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.824504942583655e-06,
+      "loss": 0.6164,
+      "step": 6250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8215682958438395e-06,
+      "loss": 0.5998,
+      "step": 6251
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.818632510645544e-06,
+      "loss": 0.5173,
+      "step": 6252
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.815697587398144e-06,
+      "loss": 0.6232,
+      "step": 6253
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8127635265108974e-06,
+      "loss": 0.868,
+      "step": 6254
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8098303283929405e-06,
+      "loss": 0.4776,
+      "step": 6255
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.806897993453287e-06,
+      "loss": 0.6193,
+      "step": 6256
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8039665221008315e-06,
+      "loss": 0.6637,
+      "step": 6257
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.801035914744352e-06,
+      "loss": 0.6314,
+      "step": 6258
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.798106171792496e-06,
+      "loss": 0.5612,
+      "step": 6259
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7951772936538e-06,
+      "loss": 0.4906,
+      "step": 6260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.792249280736677e-06,
+      "loss": 0.5417,
+      "step": 6261
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7893221334494213e-06,
+      "loss": 0.4774,
+      "step": 6262
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.786395852200199e-06,
+      "loss": 0.5801,
+      "step": 6263
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7834704373970633e-06,
+      "loss": 0.5224,
+      "step": 6264
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.780545889447943e-06,
+      "loss": 0.4654,
+      "step": 6265
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.777622208760653e-06,
+      "loss": 0.6537,
+      "step": 6266
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7746993957428725e-06,
+      "loss": 0.5217,
+      "step": 6267
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.771777450802171e-06,
+      "loss": 0.543,
+      "step": 6268
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7688563743459973e-06,
+      "loss": 0.5144,
+      "step": 6269
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7659361667816775e-06,
+      "loss": 0.8438,
+      "step": 6270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.763016828516409e-06,
+      "loss": 0.6988,
+      "step": 6271
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.760098359957275e-06,
+      "loss": 0.8538,
+      "step": 6272
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7571807615112456e-06,
+      "loss": 0.9231,
+      "step": 6273
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.754264033585151e-06,
+      "loss": 0.6359,
+      "step": 6274
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.751348176585713e-06,
+      "loss": 0.537,
+      "step": 6275
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.748433190919529e-06,
+      "loss": 0.58,
+      "step": 6276
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.745519076993078e-06,
+      "loss": 0.5952,
+      "step": 6277
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.742605835212707e-06,
+      "loss": 0.7934,
+      "step": 6278
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7396934659846507e-06,
+      "loss": 0.5209,
+      "step": 6279
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.736781969715021e-06,
+      "loss": 0.5459,
+      "step": 6280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7338713468098096e-06,
+      "loss": 0.6526,
+      "step": 6281
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.730961597674877e-06,
+      "loss": 0.5691,
+      "step": 6282
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7280527227159726e-06,
+      "loss": 0.7155,
+      "step": 6283
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7251447223387183e-06,
+      "loss": 0.6101,
+      "step": 6284
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7222375969486203e-06,
+      "loss": 0.6015,
+      "step": 6285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7193313469510504e-06,
+      "loss": 0.5623,
+      "step": 6286
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7164259727512707e-06,
+      "loss": 0.5269,
+      "step": 6287
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7135214747544144e-06,
+      "loss": 0.8259,
+      "step": 6288
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7106178533655e-06,
+      "loss": 0.674,
+      "step": 6289
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7077151089894114e-06,
+      "loss": 0.5916,
+      "step": 6290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7048132420309146e-06,
+      "loss": 0.5127,
+      "step": 6291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.701912252894666e-06,
+      "loss": 0.6445,
+      "step": 6292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6990121419851865e-06,
+      "loss": 0.6348,
+      "step": 6293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6961129097068725e-06,
+      "loss": 0.547,
+      "step": 6294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6932145564640067e-06,
+      "loss": 0.5878,
+      "step": 6295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6903170826607424e-06,
+      "loss": 0.5988,
+      "step": 6296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.687420488701119e-06,
+      "loss": 0.5657,
+      "step": 6297
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.68452477498904e-06,
+      "loss": 0.6757,
+      "step": 6298
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6816299419282974e-06,
+      "loss": 0.8591,
+      "step": 6299
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.678735989922556e-06,
+      "loss": 0.8498,
+      "step": 6300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6758429193753618e-06,
+      "loss": 0.7367,
+      "step": 6301
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.672950730690128e-06,
+      "loss": 0.644,
+      "step": 6302
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.670059424270155e-06,
+      "loss": 0.6189,
+      "step": 6303
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6671690005186146e-06,
+      "loss": 0.6073,
+      "step": 6304
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.664279459838562e-06,
+      "loss": 0.7539,
+      "step": 6305
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.661390802632918e-06,
+      "loss": 0.6295,
+      "step": 6306
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.658503029304489e-06,
+      "loss": 0.5407,
+      "step": 6307
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6556161402559576e-06,
+      "loss": 0.5886,
+      "step": 6308
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.652730135889879e-06,
+      "loss": 0.5443,
+      "step": 6309
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6498450166086906e-06,
+      "loss": 0.9061,
+      "step": 6310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6469607828147015e-06,
+      "loss": 0.5997,
+      "step": 6311
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6440774349101014e-06,
+      "loss": 0.606,
+      "step": 6312
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6411949732969497e-06,
+      "loss": 0.7777,
+      "step": 6313
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.638313398377189e-06,
+      "loss": 0.8672,
+      "step": 6314
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.635432710552634e-06,
+      "loss": 0.4773,
+      "step": 6315
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6325529102249845e-06,
+      "loss": 0.5316,
+      "step": 6316
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6296739977958008e-06,
+      "loss": 0.9251,
+      "step": 6317
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6267959736665326e-06,
+      "loss": 0.5168,
+      "step": 6318
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.623918838238499e-06,
+      "loss": 0.4676,
+      "step": 6319
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6210425919129043e-06,
+      "loss": 0.5568,
+      "step": 6320
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6181672350908127e-06,
+      "loss": 0.4997,
+      "step": 6321
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.615292768173179e-06,
+      "loss": 0.5865,
+      "step": 6322
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6124191915608266e-06,
+      "loss": 0.4301,
+      "step": 6323
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.609546505654462e-06,
+      "loss": 0.6529,
+      "step": 6324
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.606674710854654e-06,
+      "loss": 0.857,
+      "step": 6325
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6038038075618607e-06,
+      "loss": 1.0201,
+      "step": 6326
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.600933796176409e-06,
+      "loss": 0.8569,
+      "step": 6327
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.598064677098504e-06,
+      "loss": 0.8811,
+      "step": 6328
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5951964507282242e-06,
+      "loss": 0.5807,
+      "step": 6329
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5923291174655263e-06,
+      "loss": 0.7825,
+      "step": 6330
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5894626777102436e-06,
+      "loss": 0.6121,
+      "step": 6331
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.586597131862075e-06,
+      "loss": 0.6678,
+      "step": 6332
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5837324803206064e-06,
+      "loss": 0.5651,
+      "step": 6333
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5808687234852944e-06,
+      "loss": 0.8288,
+      "step": 6334
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5780058617554736e-06,
+      "loss": 0.5867,
+      "step": 6335
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.575143895530345e-06,
+      "loss": 0.8531,
+      "step": 6336
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5722828252089935e-06,
+      "loss": 0.5107,
+      "step": 6337
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5694226511903785e-06,
+      "loss": 0.6206,
+      "step": 6338
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5665633738733328e-06,
+      "loss": 0.6123,
+      "step": 6339
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5637049936565595e-06,
+      "loss": 0.8205,
+      "step": 6340
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.560847510938643e-06,
+      "loss": 0.8184,
+      "step": 6341
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5579909261180412e-06,
+      "loss": 0.6741,
+      "step": 6342
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5551352395930883e-06,
+      "loss": 0.6291,
+      "step": 6343
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5522804517619847e-06,
+      "loss": 0.9131,
+      "step": 6344
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5494265630228154e-06,
+      "loss": 0.5038,
+      "step": 6345
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5465735737735353e-06,
+      "loss": 0.6094,
+      "step": 6346
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.543721484411976e-06,
+      "loss": 0.5204,
+      "step": 6347
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5408702953358432e-06,
+      "loss": 0.5483,
+      "step": 6348
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5380200069427153e-06,
+      "loss": 0.6721,
+      "step": 6349
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5351706196300484e-06,
+      "loss": 0.5778,
+      "step": 6350
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5323221337951663e-06,
+      "loss": 0.6179,
+      "step": 6351
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5294745498352733e-06,
+      "loss": 0.8002,
+      "step": 6352
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5266278681474474e-06,
+      "loss": 0.8022,
+      "step": 6353
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5237820891286425e-06,
+      "loss": 0.8169,
+      "step": 6354
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.520937213175677e-06,
+      "loss": 0.5435,
+      "step": 6355
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.518093240685253e-06,
+      "loss": 0.8622,
+      "step": 6356
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5152501720539434e-06,
+      "loss": 0.6579,
+      "step": 6357
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5124080076781998e-06,
+      "loss": 0.5208,
+      "step": 6358
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.509566747954336e-06,
+      "loss": 0.5292,
+      "step": 6359
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.506726393278551e-06,
+      "loss": 0.6675,
+      "step": 6360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5038869440469113e-06,
+      "loss": 0.5943,
+      "step": 6361
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5010484006553657e-06,
+      "loss": 0.7094,
+      "step": 6362
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.498210763499722e-06,
+      "loss": 0.6438,
+      "step": 6363
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.495374032975669e-06,
+      "loss": 0.4264,
+      "step": 6364
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.492538209478782e-06,
+      "loss": 0.659,
+      "step": 6365
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4897032934044872e-06,
+      "loss": 0.7997,
+      "step": 6366
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4868692851480977e-06,
+      "loss": 0.8499,
+      "step": 6367
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4840361851047986e-06,
+      "loss": 0.5101,
+      "step": 6368
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4812039936696484e-06,
+      "loss": 0.5443,
+      "step": 6369
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4783727112375722e-06,
+      "loss": 0.7781,
+      "step": 6370
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.475542338203377e-06,
+      "loss": 0.5224,
+      "step": 6371
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.472712874961739e-06,
+      "loss": 0.5582,
+      "step": 6372
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.469884321907212e-06,
+      "loss": 0.5794,
+      "step": 6373
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.467056679434212e-06,
+      "loss": 0.4514,
+      "step": 6374
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.464229947937038e-06,
+      "loss": 0.6934,
+      "step": 6375
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4614041278098607e-06,
+      "loss": 0.5968,
+      "step": 6376
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.458579219446724e-06,
+      "loss": 0.5166,
+      "step": 6377
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.455755223241536e-06,
+      "loss": 0.8267,
+      "step": 6378
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4529321395880876e-06,
+      "loss": 0.5885,
+      "step": 6379
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.450109968880041e-06,
+      "loss": 0.5986,
+      "step": 6380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.44728871151093e-06,
+      "loss": 0.6001,
+      "step": 6381
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4444683678741553e-06,
+      "loss": 0.6305,
+      "step": 6382
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.441648938362996e-06,
+      "loss": 0.7363,
+      "step": 6383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4388304233706103e-06,
+      "loss": 0.5972,
+      "step": 6384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4360128232900137e-06,
+      "loss": 0.7706,
+      "step": 6385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4331961385141034e-06,
+      "loss": 0.5469,
+      "step": 6386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.43038036943565e-06,
+      "loss": 0.5886,
+      "step": 6387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4275655164472955e-06,
+      "loss": 0.5828,
+      "step": 6388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4247515799415453e-06,
+      "loss": 0.7474,
+      "step": 6389
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4219385603107903e-06,
+      "loss": 0.5323,
+      "step": 6390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4191264579472837e-06,
+      "loss": 0.6832,
+      "step": 6391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4163152732431616e-06,
+      "loss": 0.5186,
+      "step": 6392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4135050065904173e-06,
+      "loss": 0.6064,
+      "step": 6393
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.410695658380926e-06,
+      "loss": 0.7227,
+      "step": 6394
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4078872290064334e-06,
+      "loss": 0.7861,
+      "step": 6395
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4050797188585604e-06,
+      "loss": 0.5187,
+      "step": 6396
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.402273128328789e-06,
+      "loss": 0.5385,
+      "step": 6397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.399467457808483e-06,
+      "loss": 0.8383,
+      "step": 6398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3966627076888736e-06,
+      "loss": 0.65,
+      "step": 6399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3938588783610693e-06,
+      "loss": 0.5359,
+      "step": 6400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.391055970216035e-06,
+      "loss": 0.6338,
+      "step": 6401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3882539836446284e-06,
+      "loss": 0.6827,
+      "step": 6402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.385452919037564e-06,
+      "loss": 0.7185,
+      "step": 6403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3826527767854355e-06,
+      "loss": 0.6477,
+      "step": 6404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.379853557278697e-06,
+      "loss": 0.6575,
+      "step": 6405
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3770552609076845e-06,
+      "loss": 0.5555,
+      "step": 6406
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.374257888062603e-06,
+      "loss": 0.4856,
+      "step": 6407
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.37146143913353e-06,
+      "loss": 0.708,
+      "step": 6408
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3686659145104062e-06,
+      "loss": 0.8602,
+      "step": 6409
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3658713145830515e-06,
+      "loss": 0.6623,
+      "step": 6410
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3630776397411546e-06,
+      "loss": 0.6466,
+      "step": 6411
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.360284890374278e-06,
+      "loss": 0.5647,
+      "step": 6412
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3574930668718464e-06,
+      "loss": 0.5407,
+      "step": 6413
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.354702169623163e-06,
+      "loss": 0.5145,
+      "step": 6414
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.351912199017402e-06,
+      "loss": 0.5669,
+      "step": 6415
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3491231554436086e-06,
+      "loss": 0.5774,
+      "step": 6416
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.34633503929069e-06,
+      "loss": 0.5299,
+      "step": 6417
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.343547850947434e-06,
+      "loss": 0.7294,
+      "step": 6418
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3407615908024947e-06,
+      "loss": 1.0267,
+      "step": 6419
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3379762592444e-06,
+      "loss": 0.7795,
+      "step": 6420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3351918566615437e-06,
+      "loss": 0.498,
+      "step": 6421
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3324083834421928e-06,
+      "loss": 0.7819,
+      "step": 6422
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3296258399744886e-06,
+      "loss": 0.5262,
+      "step": 6423
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3268442266464308e-06,
+      "loss": 0.8059,
+      "step": 6424
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3240635438459e-06,
+      "loss": 0.5175,
+      "step": 6425
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3212837919606455e-06,
+      "loss": 0.5293,
+      "step": 6426
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.318504971378288e-06,
+      "loss": 0.6204,
+      "step": 6427
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3157270824863096e-06,
+      "loss": 0.7434,
+      "step": 6428
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.31295012567207e-06,
+      "loss": 0.5643,
+      "step": 6429
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3101741013227997e-06,
+      "loss": 0.567,
+      "step": 6430
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3073990098255983e-06,
+      "loss": 0.6983,
+      "step": 6431
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3046248515674296e-06,
+      "loss": 0.704,
+      "step": 6432
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.301851626935133e-06,
+      "loss": 0.9231,
+      "step": 6433
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.299079336315417e-06,
+      "loss": 0.8678,
+      "step": 6434
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2963079800948637e-06,
+      "loss": 0.5803,
+      "step": 6435
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2935375586599115e-06,
+      "loss": 0.6003,
+      "step": 6436
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2907680723968825e-06,
+      "loss": 0.6883,
+      "step": 6437
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2879995216919636e-06,
+      "loss": 0.681,
+      "step": 6438
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2852319069312088e-06,
+      "loss": 0.8422,
+      "step": 6439
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2824652285005465e-06,
+      "loss": 0.5876,
+      "step": 6440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.279699486785769e-06,
+      "loss": 0.6137,
+      "step": 6441
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2769346821725454e-06,
+      "loss": 0.5357,
+      "step": 6442
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2741708150464035e-06,
+      "loss": 0.8969,
+      "step": 6443
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.271407885792749e-06,
+      "loss": 0.5252,
+      "step": 6444
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2686458947968524e-06,
+      "loss": 0.5876,
+      "step": 6445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2658848424438616e-06,
+      "loss": 0.6358,
+      "step": 6446
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2631247291187785e-06,
+      "loss": 0.4978,
+      "step": 6447
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.260365555206486e-06,
+      "loss": 0.5946,
+      "step": 6448
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.257607321091734e-06,
+      "loss": 0.5727,
+      "step": 6449
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2548500271591423e-06,
+      "loss": 0.5375,
+      "step": 6450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2520936737931896e-06,
+      "loss": 0.6572,
+      "step": 6451
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2493382613782376e-06,
+      "loss": 0.9488,
+      "step": 6452
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2465837902985074e-06,
+      "loss": 0.7981,
+      "step": 6453
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.243830260938097e-06,
+      "loss": 0.5669,
+      "step": 6454
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2410776736809603e-06,
+      "loss": 0.5476,
+      "step": 6455
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2383260289109276e-06,
+      "loss": 0.5234,
+      "step": 6456
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2355753270117072e-06,
+      "loss": 0.8748,
+      "step": 6457
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2328255683668574e-06,
+      "loss": 0.5189,
+      "step": 6458
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2300767533598167e-06,
+      "loss": 0.5086,
+      "step": 6459
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.227328882373889e-06,
+      "loss": 0.6793,
+      "step": 6460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2245819557922497e-06,
+      "loss": 0.5751,
+      "step": 6461
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.221835973997934e-06,
+      "loss": 0.5442,
+      "step": 6462
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2190909373738544e-06,
+      "loss": 0.725,
+      "step": 6463
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.216346846302787e-06,
+      "loss": 0.8467,
+      "step": 6464
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2136037011673803e-06,
+      "loss": 0.8562,
+      "step": 6465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.210861502350142e-06,
+      "loss": 0.481,
+      "step": 6466
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2081202502334573e-06,
+      "loss": 0.7918,
+      "step": 6467
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2053799451995748e-06,
+      "loss": 0.9512,
+      "step": 6468
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2026405876306156e-06,
+      "loss": 0.5222,
+      "step": 6469
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.199902177908559e-06,
+      "loss": 0.5375,
+      "step": 6470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.197164716415261e-06,
+      "loss": 0.5546,
+      "step": 6471
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1944282035324413e-06,
+      "loss": 0.9218,
+      "step": 6472
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1916926396416935e-06,
+      "loss": 0.5216,
+      "step": 6473
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.188958025124467e-06,
+      "loss": 0.6996,
+      "step": 6474
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1862243603620858e-06,
+      "loss": 1.0355,
+      "step": 6475
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.183491645735751e-06,
+      "loss": 0.7505,
+      "step": 6476
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.180759881626512e-06,
+      "loss": 0.6164,
+      "step": 6477
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.178029068415298e-06,
+      "loss": 0.5397,
+      "step": 6478
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1752992064829034e-06,
+      "loss": 0.4861,
+      "step": 6479
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1725702962099923e-06,
+      "loss": 0.4911,
+      "step": 6480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.169842337977087e-06,
+      "loss": 0.6677,
+      "step": 6481
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1671153321645877e-06,
+      "loss": 0.5093,
+      "step": 6482
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.164389279152755e-06,
+      "loss": 0.9372,
+      "step": 6483
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1616641793217238e-06,
+      "loss": 0.6878,
+      "step": 6484
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.158940033051484e-06,
+      "loss": 0.8742,
+      "step": 6485
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.156216840721904e-06,
+      "loss": 0.7229,
+      "step": 6486
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1534946027127135e-06,
+      "loss": 0.8071,
+      "step": 6487
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.150773319403516e-06,
+      "loss": 0.8694,
+      "step": 6488
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1480529911737678e-06,
+      "loss": 0.688,
+      "step": 6489
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.145333618402806e-06,
+      "loss": 0.582,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1426152014698273e-06,
+      "loss": 0.4654,
+      "step": 6491
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.139897740753901e-06,
+      "loss": 0.6596,
+      "step": 6492
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1371812366339503e-06,
+      "loss": 0.4976,
+      "step": 6493
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1344656894887827e-06,
+      "loss": 0.5457,
+      "step": 6494
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1317510996970624e-06,
+      "loss": 0.6184,
+      "step": 6495
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1290374676373157e-06,
+      "loss": 0.7492,
+      "step": 6496
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1263247936879437e-06,
+      "loss": 0.7771,
+      "step": 6497
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1236130782272113e-06,
+      "loss": 0.5194,
+      "step": 6498
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.120902321633251e-06,
+      "loss": 0.534,
+      "step": 6499
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.118192524284056e-06,
+      "loss": 0.5932,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.11548368655749e-06,
+      "loss": 0.6675,
+      "step": 6501
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1127758088312853e-06,
+      "loss": 0.5674,
+      "step": 6502
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.110068891483039e-06,
+      "loss": 0.4743,
+      "step": 6503
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1073629348902055e-06,
+      "loss": 0.6192,
+      "step": 6504
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.104657939430119e-06,
+      "loss": 0.5857,
+      "step": 6505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.101953905479971e-06,
+      "loss": 0.5169,
+      "step": 6506
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.099250833416825e-06,
+      "loss": 0.5989,
+      "step": 6507
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0965487236175986e-06,
+      "loss": 0.5769,
+      "step": 6508
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.093847576459089e-06,
+      "loss": 0.5697,
+      "step": 6509
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0911473923179515e-06,
+      "loss": 0.843,
+      "step": 6510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0884481715707103e-06,
+      "loss": 0.5098,
+      "step": 6511
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.085749914593752e-06,
+      "loss": 0.5904,
+      "step": 6512
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0830526217633327e-06,
+      "loss": 0.8166,
+      "step": 6513
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.080356293455572e-06,
+      "loss": 0.5011,
+      "step": 6514
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0776609300464565e-06,
+      "loss": 0.5303,
+      "step": 6515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0749665319118314e-06,
+      "loss": 0.7401,
+      "step": 6516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0722730994274164e-06,
+      "loss": 0.6718,
+      "step": 6517
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0695806329687925e-06,
+      "loss": 0.5506,
+      "step": 6518
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.06688913291141e-06,
+      "loss": 0.641,
+      "step": 6519
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0641985996305736e-06,
+      "loss": 0.5322,
+      "step": 6520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.061509033501464e-06,
+      "loss": 0.5216,
+      "step": 6521
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0588204348991234e-06,
+      "loss": 0.6054,
+      "step": 6522
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0561328041984626e-06,
+      "loss": 0.5765,
+      "step": 6523
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0534461417742477e-06,
+      "loss": 0.4927,
+      "step": 6524
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0507604480011187e-06,
+      "loss": 0.6752,
+      "step": 6525
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0480757232535773e-06,
+      "loss": 0.87,
+      "step": 6526
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.045391967905995e-06,
+      "loss": 0.5721,
+      "step": 6527
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.042709182332597e-06,
+      "loss": 0.5153,
+      "step": 6528
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.040027366907484e-06,
+      "loss": 0.6038,
+      "step": 6529
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.037346522004616e-06,
+      "loss": 0.5482,
+      "step": 6530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0346666479978202e-06,
+      "loss": 0.555,
+      "step": 6531
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.031987745260787e-06,
+      "loss": 0.6022,
+      "step": 6532
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0293098141670707e-06,
+      "loss": 0.7803,
+      "step": 6533
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.026632855090097e-06,
+      "loss": 0.5949,
+      "step": 6534
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0239568684031407e-06,
+      "loss": 0.489,
+      "step": 6535
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0212818544793552e-06,
+      "loss": 0.5976,
+      "step": 6536
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.018607813691754e-06,
+      "loss": 0.4936,
+      "step": 6537
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.015934746413216e-06,
+      "loss": 0.5332,
+      "step": 6538
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0132626530164777e-06,
+      "loss": 0.5652,
+      "step": 6539
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0105915338741487e-06,
+      "loss": 0.8265,
+      "step": 6540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.007921389358697e-06,
+      "loss": 0.793,
+      "step": 6541
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0052522198424618e-06,
+      "loss": 0.5339,
+      "step": 6542
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0025840256976322e-06,
+      "loss": 0.9216,
+      "step": 6543
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.999916807296277e-06,
+      "loss": 0.5545,
+      "step": 6544
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9972505650103188e-06,
+      "loss": 0.6886,
+      "step": 6545
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.994585299211553e-06,
+      "loss": 0.536,
+      "step": 6546
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9919210102716256e-06,
+      "loss": 0.683,
+      "step": 6547
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.989257698562056e-06,
+      "loss": 0.7503,
+      "step": 6548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9865953644542325e-06,
+      "loss": 0.8631,
+      "step": 6549
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.983934008319391e-06,
+      "loss": 0.5331,
+      "step": 6550
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.981273630528645e-06,
+      "loss": 0.5142,
+      "step": 6551
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9786142314529642e-06,
+      "loss": 0.7093,
+      "step": 6552
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9759558114631892e-06,
+      "loss": 0.5807,
+      "step": 6553
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9732983709300136e-06,
+      "loss": 0.8007,
+      "step": 6554
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.970641910224e-06,
+      "loss": 0.6576,
+      "step": 6555
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.967986429715577e-06,
+      "loss": 0.6236,
+      "step": 6556
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.965331929775035e-06,
+      "loss": 0.5912,
+      "step": 6557
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9626784107725214e-06,
+      "loss": 0.547,
+      "step": 6558
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9600258730780564e-06,
+      "loss": 0.5783,
+      "step": 6559
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9573743170615155e-06,
+      "loss": 0.5011,
+      "step": 6560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.954723743092647e-06,
+      "loss": 0.5741,
+      "step": 6561
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.952074151541048e-06,
+      "loss": 0.7063,
+      "step": 6562
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.94942554277619e-06,
+      "loss": 0.7098,
+      "step": 6563
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.946777917167404e-06,
+      "loss": 0.5922,
+      "step": 6564
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9441312750838867e-06,
+      "loss": 0.6052,
+      "step": 6565
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9414856168946905e-06,
+      "loss": 0.9813,
+      "step": 6566
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9388409429687315e-06,
+      "loss": 0.6446,
+      "step": 6567
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.936197253674805e-06,
+      "loss": 0.569,
+      "step": 6568
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9335545493815442e-06,
+      "loss": 0.527,
+      "step": 6569
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.930912830457461e-06,
+      "loss": 0.7116,
+      "step": 6570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9282720972709255e-06,
+      "loss": 0.7599,
+      "step": 6571
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9256323501901727e-06,
+      "loss": 0.6935,
+      "step": 6572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.922993589583294e-06,
+      "loss": 0.8736,
+      "step": 6573
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9203558158182477e-06,
+      "loss": 0.5692,
+      "step": 6574
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9177190292628554e-06,
+      "loss": 0.5634,
+      "step": 6575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.915083230284802e-06,
+      "loss": 0.5076,
+      "step": 6576
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9124484192516266e-06,
+      "loss": 0.551,
+      "step": 6577
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.90981459653074e-06,
+      "loss": 0.6607,
+      "step": 6578
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9071817624894093e-06,
+      "loss": 0.5455,
+      "step": 6579
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.90454991749477e-06,
+      "loss": 0.7687,
+      "step": 6580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9019190619138106e-06,
+      "loss": 0.5273,
+      "step": 6581
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8992891961133873e-06,
+      "loss": 0.8401,
+      "step": 6582
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.896660320460218e-06,
+      "loss": 0.678,
+      "step": 6583
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.894032435320888e-06,
+      "loss": 0.6537,
+      "step": 6584
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.891405541061826e-06,
+      "loss": 0.7721,
+      "step": 6585
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.888779638049345e-06,
+      "loss": 0.5531,
+      "step": 6586
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8861547266496115e-06,
+      "loss": 0.5536,
+      "step": 6587
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.883530807228644e-06,
+      "loss": 0.5964,
+      "step": 6588
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.880907880152334e-06,
+      "loss": 0.7105,
+      "step": 6589
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.878285945786432e-06,
+      "loss": 0.5767,
+      "step": 6590
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8756650044965527e-06,
+      "loss": 0.6652,
+      "step": 6591
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8730450566481626e-06,
+      "loss": 0.5112,
+      "step": 6592
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8704261026065994e-06,
+      "loss": 0.9006,
+      "step": 6593
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8678081427370584e-06,
+      "loss": 0.647,
+      "step": 6594
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8651911774046016e-06,
+      "loss": 0.8338,
+      "step": 6595
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8625752069741387e-06,
+      "loss": 0.4998,
+      "step": 6596
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8599602318104534e-06,
+      "loss": 0.6307,
+      "step": 6597
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.857346252278189e-06,
+      "loss": 0.5476,
+      "step": 6598
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8547332687418483e-06,
+      "loss": 0.5532,
+      "step": 6599
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8521212815657886e-06,
+      "loss": 0.5987,
+      "step": 6600
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.849510291114239e-06,
+      "loss": 0.499,
+      "step": 6601
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8469002977512826e-06,
+      "loss": 0.5319,
+      "step": 6602
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.844291301840868e-06,
+      "loss": 0.5724,
+      "step": 6603
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8416833037468018e-06,
+      "loss": 0.887,
+      "step": 6604
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.839076303832751e-06,
+      "loss": 0.515,
+      "step": 6605
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8364703024622474e-06,
+      "loss": 0.6695,
+      "step": 6606
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.833865299998677e-06,
+      "loss": 0.628,
+      "step": 6607
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.831261296805292e-06,
+      "loss": 0.5991,
+      "step": 6608
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8286582932452024e-06,
+      "loss": 0.6916,
+      "step": 6609
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.826056289681384e-06,
+      "loss": 0.4856,
+      "step": 6610
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.823455286476663e-06,
+      "loss": 0.5447,
+      "step": 6611
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.820855283993734e-06,
+      "loss": 0.5484,
+      "step": 6612
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8182562825951522e-06,
+      "loss": 0.5422,
+      "step": 6613
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.815658282643332e-06,
+      "loss": 0.5432,
+      "step": 6614
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8130612845005422e-06,
+      "loss": 0.8696,
+      "step": 6615
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.810465288528922e-06,
+      "loss": 0.5549,
+      "step": 6616
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8078702950904625e-06,
+      "loss": 0.6317,
+      "step": 6617
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8052763045470233e-06,
+      "loss": 0.5667,
+      "step": 6618
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.802683317260314e-06,
+      "loss": 0.4669,
+      "step": 6619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.80009133359191e-06,
+      "loss": 0.4717,
+      "step": 6620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7975003539032486e-06,
+      "loss": 0.8283,
+      "step": 6621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.794910378555625e-06,
+      "loss": 0.5894,
+      "step": 6622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.792321407910192e-06,
+      "loss": 0.6665,
+      "step": 6623
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7897334423279666e-06,
+      "loss": 0.703,
+      "step": 6624
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7871464821698223e-06,
+      "loss": 0.5642,
+      "step": 6625
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.784560527796497e-06,
+      "loss": 0.5728,
+      "step": 6626
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7819755795685787e-06,
+      "loss": 0.6678,
+      "step": 6627
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.779391637846525e-06,
+      "loss": 0.6534,
+      "step": 6628
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.776808702990649e-06,
+      "loss": 0.5143,
+      "step": 6629
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7742267753611273e-06,
+      "loss": 0.8162,
+      "step": 6630
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7716458553179858e-06,
+      "loss": 0.5514,
+      "step": 6631
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.769065943221121e-06,
+      "loss": 0.8577,
+      "step": 6632
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.766487039430283e-06,
+      "loss": 0.748,
+      "step": 6633
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.763909144305088e-06,
+      "loss": 0.6749,
+      "step": 6634
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7613322582049994e-06,
+      "loss": 0.6893,
+      "step": 6635
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7587563814893494e-06,
+      "loss": 0.6287,
+      "step": 6636
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7561815145173275e-06,
+      "loss": 0.8528,
+      "step": 6637
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.753607657647985e-06,
+      "loss": 0.972,
+      "step": 6638
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.751034811240223e-06,
+      "loss": 0.8215,
+      "step": 6639
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7484629756528092e-06,
+      "loss": 0.6052,
+      "step": 6640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.745892151244376e-06,
+      "loss": 0.5726,
+      "step": 6641
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7433223383733996e-06,
+      "loss": 0.57,
+      "step": 6642
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7407535373982275e-06,
+      "loss": 0.4771,
+      "step": 6643
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7381857486770612e-06,
+      "loss": 0.5712,
+      "step": 6644
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.735618972567965e-06,
+      "loss": 0.5983,
+      "step": 6645
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.733053209428853e-06,
+      "loss": 0.5643,
+      "step": 6646
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7304884596175075e-06,
+      "loss": 0.5655,
+      "step": 6647
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.727924723491564e-06,
+      "loss": 0.5615,
+      "step": 6648
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7253620014085235e-06,
+      "loss": 0.6919,
+      "step": 6649
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7228002937257337e-06,
+      "loss": 0.642,
+      "step": 6650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7202396008004104e-06,
+      "loss": 0.5543,
+      "step": 6651
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7176799229896276e-06,
+      "loss": 0.6481,
+      "step": 6652
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7151212606503164e-06,
+      "loss": 0.7155,
+      "step": 6653
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.71256361413926e-06,
+      "loss": 0.6062,
+      "step": 6654
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7100069838131083e-06,
+      "loss": 0.8059,
+      "step": 6655
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7074513700283658e-06,
+      "loss": 0.6595,
+      "step": 6656
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7048967731414e-06,
+      "loss": 0.7087,
+      "step": 6657
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7023431935084267e-06,
+      "loss": 0.674,
+      "step": 6658
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6997906314855248e-06,
+      "loss": 0.5471,
+      "step": 6659
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6972390874286412e-06,
+      "loss": 0.7657,
+      "step": 6660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6946885616935647e-06,
+      "loss": 0.8566,
+      "step": 6661
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6921390546359507e-06,
+      "loss": 0.5906,
+      "step": 6662
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.68959056661131e-06,
+      "loss": 0.5824,
+      "step": 6663
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6870430979750186e-06,
+      "loss": 0.5152,
+      "step": 6664
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6844966490822942e-06,
+      "loss": 0.571,
+      "step": 6665
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.681951220288229e-06,
+      "loss": 0.6117,
+      "step": 6666
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6794068119477646e-06,
+      "loss": 0.6686,
+      "step": 6667
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6768634244157056e-06,
+      "loss": 0.5047,
+      "step": 6668
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6743210580467028e-06,
+      "loss": 0.5319,
+      "step": 6669
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.671779713195277e-06,
+      "loss": 0.6231,
+      "step": 6670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.669239390215802e-06,
+      "loss": 0.6632,
+      "step": 6671
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6667000894625118e-06,
+      "loss": 0.6344,
+      "step": 6672
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6641618112894895e-06,
+      "loss": 0.8847,
+      "step": 6673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6616245560506824e-06,
+      "loss": 0.4988,
+      "step": 6674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6590883240998966e-06,
+      "loss": 0.6176,
+      "step": 6675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.656553115790794e-06,
+      "loss": 0.8798,
+      "step": 6676
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.654018931476885e-06,
+      "loss": 0.6848,
+      "step": 6677
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.651485771511553e-06,
+      "loss": 0.5851,
+      "step": 6678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6489536362480306e-06,
+      "loss": 0.6903,
+      "step": 6679
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.646422526039402e-06,
+      "loss": 0.6128,
+      "step": 6680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6438924412386145e-06,
+      "loss": 0.6378,
+      "step": 6681
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6413633821984752e-06,
+      "loss": 0.7881,
+      "step": 6682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.638835349271646e-06,
+      "loss": 0.4967,
+      "step": 6683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6363083428106373e-06,
+      "loss": 0.7182,
+      "step": 6684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.633782363167827e-06,
+      "loss": 0.696,
+      "step": 6685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.631257410695448e-06,
+      "loss": 0.5238,
+      "step": 6686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6287334857455903e-06,
+      "loss": 0.5829,
+      "step": 6687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6262105886701918e-06,
+      "loss": 0.6864,
+      "step": 6688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.623688719821057e-06,
+      "loss": 0.5646,
+      "step": 6689
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.621167879549845e-06,
+      "loss": 0.4733,
+      "step": 6690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.618648068208073e-06,
+      "loss": 0.5454,
+      "step": 6691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.616129286147104e-06,
+      "loss": 0.594,
+      "step": 6692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.613611533718172e-06,
+      "loss": 0.7102,
+      "step": 6693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6110948112723577e-06,
+      "loss": 0.5264,
+      "step": 6694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.608579119160607e-06,
+      "loss": 0.636,
+      "step": 6695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.606064457733706e-06,
+      "loss": 0.5993,
+      "step": 6696
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6035508273423173e-06,
+      "loss": 0.6139,
+      "step": 6697
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6010382283369496e-06,
+      "loss": 0.7223,
+      "step": 6698
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5985266610679628e-06,
+      "loss": 0.7176,
+      "step": 6699
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5960161258855807e-06,
+      "loss": 0.5852,
+      "step": 6700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.593506623139881e-06,
+      "loss": 0.5536,
+      "step": 6701
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5909981531808015e-06,
+      "loss": 0.557,
+      "step": 6702
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.588490716358124e-06,
+      "loss": 0.7627,
+      "step": 6703
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5859843130214977e-06,
+      "loss": 0.6604,
+      "step": 6704
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5834789435204245e-06,
+      "loss": 0.5089,
+      "step": 6705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5809746082042643e-06,
+      "loss": 0.5066,
+      "step": 6706
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5784713074222245e-06,
+      "loss": 0.8292,
+      "step": 6707
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.575969041523375e-06,
+      "loss": 0.4956,
+      "step": 6708
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5734678108566437e-06,
+      "loss": 0.7572,
+      "step": 6709
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.570967615770811e-06,
+      "loss": 0.7594,
+      "step": 6710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5684684566145067e-06,
+      "loss": 0.5307,
+      "step": 6711
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565970333736226e-06,
+      "loss": 0.7874,
+      "step": 6712
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5634732474843162e-06,
+      "loss": 0.7592,
+      "step": 6713
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5609771982069774e-06,
+      "loss": 0.5248,
+      "step": 6714
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.558482186252269e-06,
+      "loss": 0.598,
+      "step": 6715
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5559882119681035e-06,
+      "loss": 0.4768,
+      "step": 6716
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5534952757022514e-06,
+      "loss": 0.7785,
+      "step": 6717
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5510033778023313e-06,
+      "loss": 0.5495,
+      "step": 6718
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5485125186158244e-06,
+      "loss": 0.5461,
+      "step": 6719
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5460226984900647e-06,
+      "loss": 0.6554,
+      "step": 6720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.543533917772245e-06,
+      "loss": 0.6748,
+      "step": 6721
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5410461768094007e-06,
+      "loss": 0.6041,
+      "step": 6722
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5385594759484365e-06,
+      "loss": 0.499,
+      "step": 6723
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5360738155361064e-06,
+      "loss": 0.6888,
+      "step": 6724
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5335891959190206e-06,
+      "loss": 0.6191,
+      "step": 6725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.531105617443638e-06,
+      "loss": 0.6619,
+      "step": 6726
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5286230804562794e-06,
+      "loss": 0.5805,
+      "step": 6727
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.526141585303119e-06,
+      "loss": 0.5657,
+      "step": 6728
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5236611323301884e-06,
+      "loss": 0.5822,
+      "step": 6729
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5211817218833632e-06,
+      "loss": 0.6765,
+      "step": 6730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.518703354308385e-06,
+      "loss": 0.8923,
+      "step": 6731
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5162260299508445e-06,
+      "loss": 0.5036,
+      "step": 6732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.513749749156189e-06,
+      "loss": 0.5426,
+      "step": 6733
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.511274512269719e-06,
+      "loss": 0.6051,
+      "step": 6734
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5088003196365906e-06,
+      "loss": 0.6913,
+      "step": 6735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5063271716018165e-06,
+      "loss": 0.5355,
+      "step": 6736
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.503855068510256e-06,
+      "loss": 0.6284,
+      "step": 6737
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.501384010706628e-06,
+      "loss": 0.5407,
+      "step": 6738
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.498913998535506e-06,
+      "loss": 0.7266,
+      "step": 6739
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4964450323413215e-06,
+      "loss": 0.574,
+      "step": 6740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4939771124683475e-06,
+      "loss": 0.6363,
+      "step": 6741
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4915102392607237e-06,
+      "loss": 0.5262,
+      "step": 6742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.489044413062438e-06,
+      "loss": 0.6224,
+      "step": 6743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4865796342173343e-06,
+      "loss": 0.8108,
+      "step": 6744
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.484115903069113e-06,
+      "loss": 0.9035,
+      "step": 6745
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.48165321996132e-06,
+      "loss": 0.8977,
+      "step": 6746
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4791915852373604e-06,
+      "loss": 0.8675,
+      "step": 6747
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4767309992404954e-06,
+      "loss": 0.5517,
+      "step": 6748
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4742714623138386e-06,
+      "loss": 0.7921,
+      "step": 6749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4718129748003526e-06,
+      "loss": 0.6731,
+      "step": 6750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4693555370428548e-06,
+      "loss": 0.703,
+      "step": 6751
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.466899149384029e-06,
+      "loss": 0.6353,
+      "step": 6752
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.464443812166394e-06,
+      "loss": 0.6003,
+      "step": 6753
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.461989525732331e-06,
+      "loss": 0.686,
+      "step": 6754
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.459536290424076e-06,
+      "loss": 0.5751,
+      "step": 6755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.457084106583719e-06,
+      "loss": 0.5809,
+      "step": 6756
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.454632974553195e-06,
+      "loss": 0.6612,
+      "step": 6757
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4521828946742996e-06,
+      "loss": 0.8904,
+      "step": 6758
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.449733867288683e-06,
+      "loss": 0.6337,
+      "step": 6759
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4472858927378473e-06,
+      "loss": 0.4728,
+      "step": 6760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4448389713631394e-06,
+      "loss": 0.5802,
+      "step": 6761
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.442393103505771e-06,
+      "loss": 0.6595,
+      "step": 6762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4399482895068025e-06,
+      "loss": 0.5675,
+      "step": 6763
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.43750452970715e-06,
+      "loss": 0.5978,
+      "step": 6764
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.435061824447571e-06,
+      "loss": 0.6342,
+      "step": 6765
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4326201740686905e-06,
+      "loss": 0.6489,
+      "step": 6766
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4301795789109804e-06,
+      "loss": 0.8413,
+      "step": 6767
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4277400393147665e-06,
+      "loss": 0.7151,
+      "step": 6768
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4253015556202197e-06,
+      "loss": 0.5947,
+      "step": 6769
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4228641281673792e-06,
+      "loss": 0.5586,
+      "step": 6770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4204277572961265e-06,
+      "loss": 0.5645,
+      "step": 6771
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.417992443346193e-06,
+      "loss": 0.7712,
+      "step": 6772
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4155581866571696e-06,
+      "loss": 0.645,
+      "step": 6773
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.413124987568497e-06,
+      "loss": 0.4747,
+      "step": 6774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4106928464194725e-06,
+      "loss": 0.6065,
+      "step": 6775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4082617635492354e-06,
+      "loss": 0.9344,
+      "step": 6776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.405831739296788e-06,
+      "loss": 0.5786,
+      "step": 6777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4034027740009803e-06,
+      "loss": 0.6037,
+      "step": 6778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4009748680005195e-06,
+      "loss": 0.6171,
+      "step": 6779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.398548021633954e-06,
+      "loss": 0.6404,
+      "step": 6780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.396122235239694e-06,
+      "loss": 0.6893,
+      "step": 6781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.393697509156002e-06,
+      "loss": 0.9304,
+      "step": 6782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.39127384372099e-06,
+      "loss": 0.8198,
+      "step": 6783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3888512392726183e-06,
+      "loss": 0.6151,
+      "step": 6784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3864296961487054e-06,
+      "loss": 0.5847,
+      "step": 6785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.384009214686919e-06,
+      "loss": 0.4799,
+      "step": 6786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3815897952247824e-06,
+      "loss": 0.5209,
+      "step": 6787
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.37917143809966e-06,
+      "loss": 0.6414,
+      "step": 6788
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.376754143648783e-06,
+      "loss": 0.5635,
+      "step": 6789
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.374337912209228e-06,
+      "loss": 0.5591,
+      "step": 6790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.371922744117916e-06,
+      "loss": 0.5621,
+      "step": 6791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3695086397116305e-06,
+      "loss": 0.4877,
+      "step": 6792
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3670955993270017e-06,
+      "loss": 0.9352,
+      "step": 6793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3646836233005133e-06,
+      "loss": 0.6314,
+      "step": 6794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3622727119684962e-06,
+      "loss": 0.5743,
+      "step": 6795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3598628656671383e-06,
+      "loss": 0.5557,
+      "step": 6796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3574540847324766e-06,
+      "loss": 0.5572,
+      "step": 6797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3550463695004024e-06,
+      "loss": 0.9453,
+      "step": 6798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.35263972030665e-06,
+      "loss": 0.7808,
+      "step": 6799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.350234137486813e-06,
+      "loss": 0.5246,
+      "step": 6800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.347829621376335e-06,
+      "loss": 0.5482,
+      "step": 6801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.345426172310513e-06,
+      "loss": 0.6698,
+      "step": 6802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3430237906244856e-06,
+      "loss": 0.618,
+      "step": 6803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3406224766532516e-06,
+      "loss": 0.7682,
+      "step": 6804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3382222307316593e-06,
+      "loss": 0.8069,
+      "step": 6805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.335823053194407e-06,
+      "loss": 0.6761,
+      "step": 6806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.333424944376044e-06,
+      "loss": 0.5035,
+      "step": 6807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3310279046109706e-06,
+      "loss": 0.5137,
+      "step": 6808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3286319342334406e-06,
+      "loss": 0.6641,
+      "step": 6809
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3262370335775518e-06,
+      "loss": 0.7122,
+      "step": 6810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3238432029772595e-06,
+      "loss": 0.636,
+      "step": 6811
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3214504427663676e-06,
+      "loss": 0.5964,
+      "step": 6812
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3190587532785325e-06,
+      "loss": 0.5021,
+      "step": 6813
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3166681348472553e-06,
+      "loss": 0.6414,
+      "step": 6814
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.314278587805895e-06,
+      "loss": 0.6943,
+      "step": 6815
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3118901124876567e-06,
+      "loss": 0.5678,
+      "step": 6816
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3095027092256018e-06,
+      "loss": 0.5273,
+      "step": 6817
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3071163783526306e-06,
+      "loss": 0.741,
+      "step": 6818
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3047311202015053e-06,
+      "loss": 0.553,
+      "step": 6819
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3023469351048344e-06,
+      "loss": 0.6571,
+      "step": 6820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.299963823395079e-06,
+      "loss": 0.4537,
+      "step": 6821
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2975817854045435e-06,
+      "loss": 0.5205,
+      "step": 6822
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.29520082146539e-06,
+      "loss": 0.6477,
+      "step": 6823
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2928209319096262e-06,
+      "loss": 0.5809,
+      "step": 6824
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.290442117069115e-06,
+      "loss": 0.6354,
+      "step": 6825
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2880643772755652e-06,
+      "loss": 0.5673,
+      "step": 6826
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.285687712860536e-06,
+      "loss": 0.6217,
+      "step": 6827
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2833121241554413e-06,
+      "loss": 0.6184,
+      "step": 6828
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.280937611491536e-06,
+      "loss": 0.6325,
+      "step": 6829
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2785641751999322e-06,
+      "loss": 0.6668,
+      "step": 6830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2761918156115904e-06,
+      "loss": 0.5913,
+      "step": 6831
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.273820533057324e-06,
+      "loss": 0.6293,
+      "step": 6832
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2714503278677845e-06,
+      "loss": 0.7052,
+      "step": 6833
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2690812003734873e-06,
+      "loss": 0.6785,
+      "step": 6834
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2667131509047903e-06,
+      "loss": 0.6021,
+      "step": 6835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.264346179791904e-06,
+      "loss": 0.5299,
+      "step": 6836
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2619802873648834e-06,
+      "loss": 0.5124,
+      "step": 6837
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.259615473953638e-06,
+      "loss": 0.6895,
+      "step": 6838
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2572517398879255e-06,
+      "loss": 0.636,
+      "step": 6839
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.254889085497356e-06,
+      "loss": 0.6367,
+      "step": 6840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.252527511111381e-06,
+      "loss": 0.8197,
+      "step": 6841
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2501670170593094e-06,
+      "loss": 0.551,
+      "step": 6842
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2478076036702958e-06,
+      "loss": 1.017,
+      "step": 6843
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2454492712733446e-06,
+      "loss": 0.829,
+      "step": 6844
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2430920201973096e-06,
+      "loss": 0.7803,
+      "step": 6845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2407358507708942e-06,
+      "loss": 0.787,
+      "step": 6846
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.238380763322655e-06,
+      "loss": 0.6833,
+      "step": 6847
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2360267581809857e-06,
+      "loss": 0.6143,
+      "step": 6848
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2336738356741403e-06,
+      "loss": 0.7712,
+      "step": 6849
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.231321996130218e-06,
+      "loss": 0.533,
+      "step": 6850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2289712398771714e-06,
+      "loss": 0.5367,
+      "step": 6851
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2266215672427905e-06,
+      "loss": 0.5468,
+      "step": 6852
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2242729785547257e-06,
+      "loss": 0.6079,
+      "step": 6853
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2219254741404717e-06,
+      "loss": 0.635,
+      "step": 6854
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2195790543273756e-06,
+      "loss": 0.6979,
+      "step": 6855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.217233719442624e-06,
+      "loss": 0.5103,
+      "step": 6856
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2148894698132617e-06,
+      "loss": 0.5416,
+      "step": 6857
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2125463057661787e-06,
+      "loss": 0.7756,
+      "step": 6858
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2102042276281167e-06,
+      "loss": 0.6476,
+      "step": 6859
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2078632357256567e-06,
+      "loss": 0.6279,
+      "step": 6860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2055233303852353e-06,
+      "loss": 0.6319,
+      "step": 6861
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.203184511933143e-06,
+      "loss": 0.5117,
+      "step": 6862
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2008467806955115e-06,
+      "loss": 0.7106,
+      "step": 6863
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1985101369983166e-06,
+      "loss": 0.6822,
+      "step": 6864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1961745811673905e-06,
+      "loss": 0.6039,
+      "step": 6865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.193840113528413e-06,
+      "loss": 0.7898,
+      "step": 6866
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1915067344069097e-06,
+      "loss": 0.566,
+      "step": 6867
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1891744441282516e-06,
+      "loss": 0.9387,
+      "step": 6868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1868432430176634e-06,
+      "loss": 0.5278,
+      "step": 6869
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1845131314002153e-06,
+      "loss": 0.5809,
+      "step": 6870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1821841096008287e-06,
+      "loss": 0.69,
+      "step": 6871
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1798561779442662e-06,
+      "loss": 0.5668,
+      "step": 6872
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1775293367551443e-06,
+      "loss": 0.5129,
+      "step": 6873
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1752035863579257e-06,
+      "loss": 0.6214,
+      "step": 6874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1728789270769236e-06,
+      "loss": 0.6527,
+      "step": 6875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1705553592362904e-06,
+      "loss": 0.4965,
+      "step": 6876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1682328831600352e-06,
+      "loss": 0.6286,
+      "step": 6877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.165911499172013e-06,
+      "loss": 0.4462,
+      "step": 6878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1635912075959274e-06,
+      "loss": 0.6364,
+      "step": 6879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.161272008755321e-06,
+      "loss": 0.4909,
+      "step": 6880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.158953902973596e-06,
+      "loss": 0.5006,
+      "step": 6881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1566368905739987e-06,
+      "loss": 0.5472,
+      "step": 6882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1543209718796164e-06,
+      "loss": 0.6346,
+      "step": 6883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1520061472133903e-06,
+      "loss": 0.5464,
+      "step": 6884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.149692416898107e-06,
+      "loss": 0.5788,
+      "step": 6885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1473797812564056e-06,
+      "loss": 0.553,
+      "step": 6886
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1450682406107603e-06,
+      "loss": 0.4808,
+      "step": 6887
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1427577952835044e-06,
+      "loss": 0.5749,
+      "step": 6888
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1404484455968145e-06,
+      "loss": 0.5191,
+      "step": 6889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.138140191872715e-06,
+      "loss": 0.4954,
+      "step": 6890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1358330344330724e-06,
+      "loss": 0.5982,
+      "step": 6891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.133526973599607e-06,
+      "loss": 0.5921,
+      "step": 6892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.131222009693884e-06,
+      "loss": 0.8361,
+      "step": 6893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1289181430373185e-06,
+      "loss": 0.8468,
+      "step": 6894
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1266153739511632e-06,
+      "loss": 0.6179,
+      "step": 6895
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1243137027565276e-06,
+      "loss": 0.6101,
+      "step": 6896
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.122013129774364e-06,
+      "loss": 0.6943,
+      "step": 6897
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1197136553254727e-06,
+      "loss": 0.502,
+      "step": 6898
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1174152797304993e-06,
+      "loss": 0.6303,
+      "step": 6899
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.115118003309937e-06,
+      "loss": 0.8186,
+      "step": 6900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1128218263841304e-06,
+      "loss": 0.5074,
+      "step": 6901
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1105267492732595e-06,
+      "loss": 0.5971,
+      "step": 6902
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1082327722973593e-06,
+      "loss": 0.6434,
+      "step": 6903
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.105939895776312e-06,
+      "loss": 0.7954,
+      "step": 6904
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1036481200298453e-06,
+      "loss": 0.5932,
+      "step": 6905
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1013574453775264e-06,
+      "loss": 0.6592,
+      "step": 6906
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.099067872138779e-06,
+      "loss": 0.6294,
+      "step": 6907
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.096779400632869e-06,
+      "loss": 0.569,
+      "step": 6908
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0944920311789095e-06,
+      "loss": 0.5551,
+      "step": 6909
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0922057640958547e-06,
+      "loss": 0.5232,
+      "step": 6910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.089920599702513e-06,
+      "loss": 0.5155,
+      "step": 6911
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0876365383175335e-06,
+      "loss": 0.5271,
+      "step": 6912
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0853535802594184e-06,
+      "loss": 0.665,
+      "step": 6913
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.083071725846504e-06,
+      "loss": 0.7489,
+      "step": 6914
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0807909753969836e-06,
+      "loss": 0.7201,
+      "step": 6915
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.078511329228893e-06,
+      "loss": 0.639,
+      "step": 6916
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.076232787660112e-06,
+      "loss": 0.5064,
+      "step": 6917
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.07395535100837e-06,
+      "loss": 0.7778,
+      "step": 6918
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0716790195912394e-06,
+      "loss": 0.8979,
+      "step": 6919
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0694037937261437e-06,
+      "loss": 0.5239,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0671296737303415e-06,
+      "loss": 0.6524,
+      "step": 6921
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0648566599209475e-06,
+      "loss": 0.5086,
+      "step": 6922
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.062584752614917e-06,
+      "loss": 0.8299,
+      "step": 6923
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.060313952129056e-06,
+      "loss": 0.536,
+      "step": 6924
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0580442587800075e-06,
+      "loss": 0.4884,
+      "step": 6925
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0557756728842682e-06,
+      "loss": 0.5444,
+      "step": 6926
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.053508194758176e-06,
+      "loss": 0.5409,
+      "step": 6927
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.05124182471792e-06,
+      "loss": 0.6259,
+      "step": 6928
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0489765630795245e-06,
+      "loss": 0.6137,
+      "step": 6929
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.046712410158869e-06,
+      "loss": 0.6009,
+      "step": 6930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0444493662716723e-06,
+      "loss": 0.6321,
+      "step": 6931
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0421874317335055e-06,
+      "loss": 0.5855,
+      "step": 6932
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0399266068597743e-06,
+      "loss": 0.5876,
+      "step": 6933
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0376668919657394e-06,
+      "loss": 0.6098,
+      "step": 6934
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0354082873665015e-06,
+      "loss": 0.6166,
+      "step": 6935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.03315079337701e-06,
+      "loss": 0.5493,
+      "step": 6936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0308944103120564e-06,
+      "loss": 0.5865,
+      "step": 6937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0286391384862793e-06,
+      "loss": 0.5969,
+      "step": 6938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.026384978214163e-06,
+      "loss": 0.4923,
+      "step": 6939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0241319298100302e-06,
+      "loss": 0.5624,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0218799935880575e-06,
+      "loss": 0.6285,
+      "step": 6941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0196291698622617e-06,
+      "loss": 0.576,
+      "step": 6942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.017379458946507e-06,
+      "loss": 0.5304,
+      "step": 6943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0151308611544963e-06,
+      "loss": 0.5374,
+      "step": 6944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0128833767997847e-06,
+      "loss": 0.4939,
+      "step": 6945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.010637006195768e-06,
+      "loss": 0.5774,
+      "step": 6946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0083917496556926e-06,
+      "loss": 0.5159,
+      "step": 6947
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0061476074926366e-06,
+      "loss": 0.809,
+      "step": 6948
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0039045800195356e-06,
+      "loss": 0.4794,
+      "step": 6949
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0016626675491637e-06,
+      "loss": 0.8852,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.999421870394145e-06,
+      "loss": 0.6076,
+      "step": 6951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9971821888669364e-06,
+      "loss": 0.4973,
+      "step": 6952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.994943623279848e-06,
+      "loss": 0.5814,
+      "step": 6953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9927061739450404e-06,
+      "loss": 0.4933,
+      "step": 6954
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.990469841174505e-06,
+      "loss": 0.6019,
+      "step": 6955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9882346252800833e-06,
+      "loss": 0.7626,
+      "step": 6956
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.986000526573464e-06,
+      "loss": 0.718,
+      "step": 6957
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9837675453661786e-06,
+      "loss": 0.5307,
+      "step": 6958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9815356819695962e-06,
+      "loss": 0.8826,
+      "step": 6959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9793049366949392e-06,
+      "loss": 0.5835,
+      "step": 6960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.977075309853269e-06,
+      "loss": 0.5039,
+      "step": 6961
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9748468017554966e-06,
+      "loss": 0.5494,
+      "step": 6962
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9726194127123666e-06,
+      "loss": 0.5751,
+      "step": 6963
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.970393143034476e-06,
+      "loss": 0.5192,
+      "step": 6964
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968167993032263e-06,
+      "loss": 0.6106,
+      "step": 6965
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9659439630160136e-06,
+      "loss": 0.7737,
+      "step": 6966
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9637210532958495e-06,
+      "loss": 0.7244,
+      "step": 6967
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.961499264181742e-06,
+      "loss": 0.5555,
+      "step": 6968
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.959278595983506e-06,
+      "loss": 0.5247,
+      "step": 6969
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9570590490108012e-06,
+      "loss": 0.7154,
+      "step": 6970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.954840623573122e-06,
+      "loss": 0.7915,
+      "step": 6971
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.952623319979816e-06,
+      "loss": 0.7481,
+      "step": 6972
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.950407138540075e-06,
+      "loss": 0.5672,
+      "step": 6973
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9481920795629317e-06,
+      "loss": 0.563,
+      "step": 6974
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9459781433572557e-06,
+      "loss": 0.5033,
+      "step": 6975
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9437653302317683e-06,
+      "loss": 0.5678,
+      "step": 6976
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.941553640495032e-06,
+      "loss": 0.7829,
+      "step": 6977
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9393430744554543e-06,
+      "loss": 0.5566,
+      "step": 6978
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.937133632421281e-06,
+      "loss": 0.6294,
+      "step": 6979
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.934925314700603e-06,
+      "loss": 0.9165,
+      "step": 6980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.93271812160136e-06,
+      "loss": 0.8461,
+      "step": 6981
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9305120534313295e-06,
+      "loss": 0.6972,
+      "step": 6982
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.92830711049813e-06,
+      "loss": 0.5613,
+      "step": 6983
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9261032931092283e-06,
+      "loss": 0.5812,
+      "step": 6984
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9239006015719318e-06,
+      "loss": 0.7412,
+      "step": 6985
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.921699036193395e-06,
+      "loss": 0.6006,
+      "step": 6986
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.919498597280606e-06,
+      "loss": 0.5613,
+      "step": 6987
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9172992851404026e-06,
+      "loss": 0.5297,
+      "step": 6988
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.915101100079465e-06,
+      "loss": 0.843,
+      "step": 6989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9129040424043176e-06,
+      "loss": 0.7178,
+      "step": 6990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9107081124213235e-06,
+      "loss": 0.4432,
+      "step": 6991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9085133104366915e-06,
+      "loss": 0.5753,
+      "step": 6992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.906319636756474e-06,
+      "loss": 0.5305,
+      "step": 6993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9041270916865607e-06,
+      "loss": 0.5732,
+      "step": 6994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.901935675532689e-06,
+      "loss": 0.8566,
+      "step": 6995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.899745388600437e-06,
+      "loss": 0.5446,
+      "step": 6996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.897556231195229e-06,
+      "loss": 0.5945,
+      "step": 6997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8953682036223243e-06,
+      "loss": 0.9719,
+      "step": 6998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8931813061868287e-06,
+      "loss": 0.5352,
+      "step": 6999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.890995539193694e-06,
+      "loss": 0.5089,
+      "step": 7000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8888109029477108e-06,
+      "loss": 0.487,
+      "step": 7001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.88662739775351e-06,
+      "loss": 0.894,
+      "step": 7002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8844450239155665e-06,
+      "loss": 0.6606,
+      "step": 7003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8822637817381996e-06,
+      "loss": 0.7818,
+      "step": 7004
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.880083671525571e-06,
+      "loss": 0.7237,
+      "step": 7005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8779046935816768e-06,
+      "loss": 0.8339,
+      "step": 7006
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8757268482103662e-06,
+      "loss": 0.7162,
+      "step": 7007
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.873550135715324e-06,
+      "loss": 0.5084,
+      "step": 7008
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8713745564000774e-06,
+      "loss": 0.5115,
+      "step": 7009
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.869200110567998e-06,
+      "loss": 0.5102,
+      "step": 7010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8670267985222978e-06,
+      "loss": 0.5474,
+      "step": 7011
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8648546205660324e-06,
+      "loss": 0.6069,
+      "step": 7012
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8626835770020924e-06,
+      "loss": 0.8819,
+      "step": 7013
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8605136681332193e-06,
+      "loss": 0.4912,
+      "step": 7014
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.858344894261992e-06,
+      "loss": 0.4848,
+      "step": 7015
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8561772556908353e-06,
+      "loss": 0.5044,
+      "step": 7016
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8540107527220052e-06,
+      "loss": 0.5677,
+      "step": 7017
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.851845385657609e-06,
+      "loss": 0.5774,
+      "step": 7018
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8496811547995941e-06,
+      "loss": 0.5795,
+      "step": 7019
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8475180604497501e-06,
+      "loss": 0.582,
+      "step": 7020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8453561029097e-06,
+      "loss": 0.7459,
+      "step": 7021
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8431952824809197e-06,
+      "loss": 0.703,
+      "step": 7022
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8410355994647188e-06,
+      "loss": 0.6044,
+      "step": 7023
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8388770541622547e-06,
+      "loss": 0.6491,
+      "step": 7024
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8367196468745162e-06,
+      "loss": 0.5758,
+      "step": 7025
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8345633779023442e-06,
+      "loss": 0.4904,
+      "step": 7026
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8324082475464133e-06,
+      "loss": 0.6604,
+      "step": 7027
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8302542561072445e-06,
+      "loss": 0.6618,
+      "step": 7028
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8281014038851963e-06,
+      "loss": 0.595,
+      "step": 7029
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8259496911804709e-06,
+      "loss": 0.7766,
+      "step": 7030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8237991182931126e-06,
+      "loss": 0.6345,
+      "step": 7031
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8216496855229981e-06,
+      "loss": 0.525,
+      "step": 7032
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8195013931698557e-06,
+      "loss": 0.4688,
+      "step": 7033
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.81735424153325e-06,
+      "loss": 0.5749,
+      "step": 7034
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8152082309125908e-06,
+      "loss": 0.5161,
+      "step": 7035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.813063361607118e-06,
+      "loss": 0.7453,
+      "step": 7036
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.810919633915923e-06,
+      "loss": 0.514,
+      "step": 7037
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8087770481379352e-06,
+      "loss": 0.5662,
+      "step": 7038
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8066356045719257e-06,
+      "loss": 0.5693,
+      "step": 7039
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8044953035165002e-06,
+      "loss": 0.7604,
+      "step": 7040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8023561452701111e-06,
+      "loss": 0.8642,
+      "step": 7041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8002181301310517e-06,
+      "loss": 0.5563,
+      "step": 7042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7980812583974539e-06,
+      "loss": 0.5352,
+      "step": 7043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.795945530367288e-06,
+      "loss": 0.9221,
+      "step": 7044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7938109463383657e-06,
+      "loss": 0.8269,
+      "step": 7045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.791677506608348e-06,
+      "loss": 0.5945,
+      "step": 7046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7895452114747226e-06,
+      "loss": 0.5882,
+      "step": 7047
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7874140612348256e-06,
+      "loss": 0.708,
+      "step": 7048
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7852840561858332e-06,
+      "loss": 0.5305,
+      "step": 7049
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7831551966247607e-06,
+      "loss": 0.5534,
+      "step": 7050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7810274828484608e-06,
+      "loss": 0.6075,
+      "step": 7051
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7789009151536296e-06,
+      "loss": 0.5651,
+      "step": 7052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7767754938368042e-06,
+      "loss": 0.4707,
+      "step": 7053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7746512191943633e-06,
+      "loss": 0.6435,
+      "step": 7054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7725280915225174e-06,
+      "loss": 0.7251,
+      "step": 7055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7704061111173253e-06,
+      "loss": 0.7153,
+      "step": 7056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7682852782746829e-06,
+      "loss": 0.5425,
+      "step": 7057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7661655932903287e-06,
+      "loss": 0.8243,
+      "step": 7058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7640470564598345e-06,
+      "loss": 0.7179,
+      "step": 7059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7619296680786191e-06,
+      "loss": 0.5533,
+      "step": 7060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7598134284419366e-06,
+      "loss": 0.6151,
+      "step": 7061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7576983378448865e-06,
+      "loss": 0.5389,
+      "step": 7062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7555843965823992e-06,
+      "loss": 0.4881,
+      "step": 7063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7534716049492494e-06,
+      "loss": 0.5181,
+      "step": 7064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.751359963240059e-06,
+      "loss": 0.5529,
+      "step": 7065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.749249471749277e-06,
+      "loss": 0.5895,
+      "step": 7066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7471401307711977e-06,
+      "loss": 0.9159,
+      "step": 7067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.745031940599956e-06,
+      "loss": 0.8309,
+      "step": 7068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.742924901529527e-06,
+      "loss": 0.5799,
+      "step": 7069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7408190138537183e-06,
+      "loss": 0.5726,
+      "step": 7070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7387142778661858e-06,
+      "loss": 0.6671,
+      "step": 7071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7366106938604198e-06,
+      "loss": 0.5908,
+      "step": 7072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7345082621297548e-06,
+      "loss": 0.5701,
+      "step": 7073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.732406982967355e-06,
+      "loss": 0.5091,
+      "step": 7074
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7303068566662317e-06,
+      "loss": 0.7728,
+      "step": 7075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7282078835192362e-06,
+      "loss": 0.8117,
+      "step": 7076
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7261100638190564e-06,
+      "loss": 0.5764,
+      "step": 7077
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7240133978582162e-06,
+      "loss": 0.647,
+      "step": 7078
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.721917885929083e-06,
+      "loss": 0.5796,
+      "step": 7079
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7198235283238629e-06,
+      "loss": 0.5264,
+      "step": 7080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7177303253346022e-06,
+      "loss": 0.5618,
+      "step": 7081
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7156382772531777e-06,
+      "loss": 0.666,
+      "step": 7082
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.713547384371318e-06,
+      "loss": 0.5113,
+      "step": 7083
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7114576469805822e-06,
+      "loss": 0.7941,
+      "step": 7084
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7093690653723737e-06,
+      "loss": 0.5813,
+      "step": 7085
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7072816398379256e-06,
+      "loss": 0.836,
+      "step": 7086
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7051953706683178e-06,
+      "loss": 0.5295,
+      "step": 7087
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7031102581544679e-06,
+      "loss": 0.4968,
+      "step": 7088
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7010263025871321e-06,
+      "loss": 0.6166,
+      "step": 7089
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6989435042569013e-06,
+      "loss": 0.6031,
+      "step": 7090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.696861863454209e-06,
+      "loss": 0.7371,
+      "step": 7091
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6947813804693258e-06,
+      "loss": 0.4961,
+      "step": 7092
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6927020555923645e-06,
+      "loss": 0.7039,
+      "step": 7093
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6906238891132687e-06,
+      "loss": 0.6126,
+      "step": 7094
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6885468813218276e-06,
+      "loss": 0.6158,
+      "step": 7095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.686471032507665e-06,
+      "loss": 0.7182,
+      "step": 7096
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6843963429602484e-06,
+      "loss": 0.6111,
+      "step": 7097
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6823228129688728e-06,
+      "loss": 0.5529,
+      "step": 7098
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6802504428226819e-06,
+      "loss": 0.6468,
+      "step": 7099
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6781792328106539e-06,
+      "loss": 0.8293,
+      "step": 7100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6761091832216047e-06,
+      "loss": 0.754,
+      "step": 7101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6740402943441902e-06,
+      "loss": 0.8764,
+      "step": 7102
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6719725664669028e-06,
+      "loss": 0.5001,
+      "step": 7103
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6699059998780754e-06,
+      "loss": 0.6941,
+      "step": 7104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6678405948658737e-06,
+      "loss": 0.539,
+      "step": 7105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6657763517183046e-06,
+      "loss": 0.637,
+      "step": 7106
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6637132707232161e-06,
+      "loss": 0.4526,
+      "step": 7107
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.661651352168292e-06,
+      "loss": 0.5797,
+      "step": 7108
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6595905963410486e-06,
+      "loss": 0.6496,
+      "step": 7109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6575310035288473e-06,
+      "loss": 0.5054,
+      "step": 7110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.655472574018885e-06,
+      "loss": 0.5116,
+      "step": 7111
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6534153080981973e-06,
+      "loss": 0.6393,
+      "step": 7112
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6513592060536532e-06,
+      "loss": 0.8126,
+      "step": 7113
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6493042681719628e-06,
+      "loss": 0.6431,
+      "step": 7114
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.647250494739676e-06,
+      "loss": 0.8951,
+      "step": 7115
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6451978860431806e-06,
+      "loss": 0.5671,
+      "step": 7116
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6431464423686915e-06,
+      "loss": 0.9477,
+      "step": 7117
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6410961640022738e-06,
+      "loss": 0.6257,
+      "step": 7118
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.639047051229825e-06,
+      "loss": 0.5835,
+      "step": 7119
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6369991043370792e-06,
+      "loss": 0.6147,
+      "step": 7120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6349523236096098e-06,
+      "loss": 0.5647,
+      "step": 7121
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6329067093328278e-06,
+      "loss": 0.6403,
+      "step": 7122
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6308622617919823e-06,
+      "loss": 0.472,
+      "step": 7123
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6288189812721522e-06,
+      "loss": 0.6152,
+      "step": 7124
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.626776868058263e-06,
+      "loss": 0.6128,
+      "step": 7125
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6247359224350744e-06,
+      "loss": 0.5958,
+      "step": 7126
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6226961446871836e-06,
+      "loss": 0.5847,
+      "step": 7127
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6206575350990205e-06,
+      "loss": 0.7004,
+      "step": 7128
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.618620093954858e-06,
+      "loss": 0.6582,
+      "step": 7129
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.616583821538803e-06,
+      "loss": 0.6858,
+      "step": 7130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.614548718134804e-06,
+      "loss": 0.6024,
+      "step": 7131
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6125147840266376e-06,
+      "loss": 0.8398,
+      "step": 7132
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6104820194979232e-06,
+      "loss": 0.6683,
+      "step": 7133
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6084504248321187e-06,
+      "loss": 0.7481,
+      "step": 7134
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6064200003125186e-06,
+      "loss": 0.739,
+      "step": 7135
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6043907462222453e-06,
+      "loss": 0.5991,
+      "step": 7136
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6023626628442678e-06,
+      "loss": 0.6101,
+      "step": 7137
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6003357504613937e-06,
+      "loss": 0.5866,
+      "step": 7138
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5983100093562566e-06,
+      "loss": 0.8701,
+      "step": 7139
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5962854398113359e-06,
+      "loss": 0.5695,
+      "step": 7140
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5942620421089427e-06,
+      "loss": 0.5778,
+      "step": 7141
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5922398165312292e-06,
+      "loss": 0.5501,
+      "step": 7142
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5902187633601773e-06,
+      "loss": 0.6197,
+      "step": 7143
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5881988828776118e-06,
+      "loss": 0.5256,
+      "step": 7144
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5861801753651918e-06,
+      "loss": 0.603,
+      "step": 7145
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5841626411044154e-06,
+      "loss": 0.6024,
+      "step": 7146
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5821462803766087e-06,
+      "loss": 0.5127,
+      "step": 7147
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5801310934629422e-06,
+      "loss": 0.5126,
+      "step": 7148
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5781170806444213e-06,
+      "loss": 0.5699,
+      "step": 7149
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.576104242201888e-06,
+      "loss": 0.8012,
+      "step": 7150
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5740925784160165e-06,
+      "loss": 0.5403,
+      "step": 7151
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5720820895673216e-06,
+      "loss": 0.7595,
+      "step": 7152
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5700727759361512e-06,
+      "loss": 0.486,
+      "step": 7153
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5680646378026954e-06,
+      "loss": 0.6789,
+      "step": 7154
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5660576754469691e-06,
+      "loss": 0.6129,
+      "step": 7155
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564051889148832e-06,
+      "loss": 0.8571,
+      "step": 7156
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5620472791879836e-06,
+      "loss": 0.7434,
+      "step": 7157
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.560043845843946e-06,
+      "loss": 0.5919,
+      "step": 7158
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5580415893960886e-06,
+      "loss": 0.5671,
+      "step": 7159
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.556040510123612e-06,
+      "loss": 0.5483,
+      "step": 7160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5540406083055548e-06,
+      "loss": 0.5495,
+      "step": 7161
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5520418842207874e-06,
+      "loss": 0.6684,
+      "step": 7162
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5500443381480202e-06,
+      "loss": 0.8021,
+      "step": 7163
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5480479703657969e-06,
+      "loss": 0.6624,
+      "step": 7164
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5460527811525005e-06,
+      "loss": 0.7259,
+      "step": 7165
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5440587707863442e-06,
+      "loss": 0.7635,
+      "step": 7166
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5420659395453796e-06,
+      "loss": 0.6791,
+      "step": 7167
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5400742877074947e-06,
+      "loss": 0.6635,
+      "step": 7168
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5380838155504151e-06,
+      "loss": 0.6449,
+      "step": 7169
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5360945233516933e-06,
+      "loss": 0.6521,
+      "step": 7170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.534106411388726e-06,
+      "loss": 0.5827,
+      "step": 7171
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5321194799387428e-06,
+      "loss": 0.8089,
+      "step": 7172
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5301337292788088e-06,
+      "loss": 0.7396,
+      "step": 7173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5281491596858178e-06,
+      "loss": 0.753,
+      "step": 7174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.526165771436513e-06,
+      "loss": 0.5501,
+      "step": 7175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5241835648074622e-06,
+      "loss": 0.5423,
+      "step": 7176
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5222025400750695e-06,
+      "loss": 0.6782,
+      "step": 7177
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5202226975155753e-06,
+      "loss": 0.6591,
+      "step": 7178
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5182440374050578e-06,
+      "loss": 0.5948,
+      "step": 7179
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5162665600194282e-06,
+      "loss": 0.4653,
+      "step": 7180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5142902656344294e-06,
+      "loss": 0.6705,
+      "step": 7181
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5123151545256454e-06,
+      "loss": 0.8088,
+      "step": 7182
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5103412269684913e-06,
+      "loss": 0.6111,
+      "step": 7183
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5083684832382216e-06,
+      "loss": 0.6461,
+      "step": 7184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5063969236099163e-06,
+      "loss": 0.6007,
+      "step": 7185
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5044265483584996e-06,
+      "loss": 0.6017,
+      "step": 7186
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5024573577587277e-06,
+      "loss": 0.6353,
+      "step": 7187
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5004893520851937e-06,
+      "loss": 0.6644,
+      "step": 7188
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4985225316123176e-06,
+      "loss": 0.7845,
+      "step": 7189
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4965568966143607e-06,
+      "loss": 0.6473,
+      "step": 7190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4945924473654205e-06,
+      "loss": 0.5469,
+      "step": 7191
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.492629184139427e-06,
+      "loss": 0.6019,
+      "step": 7192
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4906671072101387e-06,
+      "loss": 0.6522,
+      "step": 7193
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.488706216851159e-06,
+      "loss": 0.5908,
+      "step": 7194
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4867465133359216e-06,
+      "loss": 0.8113,
+      "step": 7195
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4847879969376932e-06,
+      "loss": 0.5731,
+      "step": 7196
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4828306679295746e-06,
+      "loss": 0.6766,
+      "step": 7197
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4808745265845038e-06,
+      "loss": 0.5664,
+      "step": 7198
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4789195731752504e-06,
+      "loss": 0.648,
+      "step": 7199
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.476965807974423e-06,
+      "loss": 0.632,
+      "step": 7200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4750132312544575e-06,
+      "loss": 0.5339,
+      "step": 7201
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4730618432876288e-06,
+      "loss": 0.5618,
+      "step": 7202
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.471111644346046e-06,
+      "loss": 0.7362,
+      "step": 7203
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.469162634701653e-06,
+      "loss": 0.6157,
+      "step": 7204
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4672148146262221e-06,
+      "loss": 0.7644,
+      "step": 7205
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4652681843913662e-06,
+      "loss": 0.8336,
+      "step": 7206
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4633227442685306e-06,
+      "loss": 0.8156,
+      "step": 7207
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4613784945289956e-06,
+      "loss": 0.461,
+      "step": 7208
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4594354354438688e-06,
+      "loss": 0.7666,
+      "step": 7209
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4574935672841017e-06,
+      "loss": 0.6454,
+      "step": 7210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4555528903204718e-06,
+      "loss": 0.5178,
+      "step": 7211
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.453613404823596e-06,
+      "loss": 0.6542,
+      "step": 7212
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4516751110639227e-06,
+      "loss": 0.642,
+      "step": 7213
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.449738009311733e-06,
+      "loss": 0.5798,
+      "step": 7214
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4478020998371456e-06,
+      "loss": 0.5768,
+      "step": 7215
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4458673829101067e-06,
+      "loss": 0.7965,
+      "step": 7216
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4439338588004005e-06,
+      "loss": 0.6226,
+      "step": 7217
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4420015277776455e-06,
+      "loss": 0.5165,
+      "step": 7218
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4400703901112946e-06,
+      "loss": 0.4346,
+      "step": 7219
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.438140446070626e-06,
+      "loss": 0.5705,
+      "step": 7220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4362116959247619e-06,
+      "loss": 0.5318,
+      "step": 7221
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4342841399426532e-06,
+      "loss": 0.6734,
+      "step": 7222
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4323577783930876e-06,
+      "loss": 0.482,
+      "step": 7223
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4304326115446777e-06,
+      "loss": 0.8615,
+      "step": 7224
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4285086396658787e-06,
+      "loss": 0.7416,
+      "step": 7225
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.426585863024975e-06,
+      "loss": 0.5809,
+      "step": 7226
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.424664281890088e-06,
+      "loss": 0.5821,
+      "step": 7227
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4227438965291651e-06,
+      "loss": 0.5199,
+      "step": 7228
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.42082470720999e-06,
+      "loss": 0.4645,
+      "step": 7229
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.41890671420019e-06,
+      "loss": 0.6278,
+      "step": 7230
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4169899177672076e-06,
+      "loss": 0.6092,
+      "step": 7231
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4150743181783312e-06,
+      "loss": 0.6002,
+      "step": 7232
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.413159915700677e-06,
+      "loss": 0.5335,
+      "step": 7233
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4112467106012007e-06,
+      "loss": 0.8902,
+      "step": 7234
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4093347031466786e-06,
+      "loss": 0.6302,
+      "step": 7235
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.407423893603731e-06,
+      "loss": 0.4772,
+      "step": 7236
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4055142822388079e-06,
+      "loss": 0.4774,
+      "step": 7237
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4036058693181943e-06,
+      "loss": 0.5584,
+      "step": 7238
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4016986551080013e-06,
+      "loss": 0.6395,
+      "step": 7239
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.399792639874179e-06,
+      "loss": 0.6096,
+      "step": 7240
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3978878238825088e-06,
+      "loss": 0.5817,
+      "step": 7241
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3959842073986085e-06,
+      "loss": 0.5523,
+      "step": 7242
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3940817906879177e-06,
+      "loss": 0.5009,
+      "step": 7243
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3921805740157201e-06,
+      "loss": 0.5721,
+      "step": 7244
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3902805576471268e-06,
+      "loss": 0.5948,
+      "step": 7245
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3883817418470847e-06,
+      "loss": 0.4978,
+      "step": 7246
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3864841268803675e-06,
+      "loss": 0.6377,
+      "step": 7247
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3845877130115847e-06,
+      "loss": 0.9163,
+      "step": 7248
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3826925005051861e-06,
+      "loss": 0.5959,
+      "step": 7249
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3807984896254389e-06,
+      "loss": 0.5927,
+      "step": 7250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.378905680636452e-06,
+      "loss": 0.8471,
+      "step": 7251
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3770140738021665e-06,
+      "loss": 0.7313,
+      "step": 7252
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3751236693863567e-06,
+      "loss": 0.4768,
+      "step": 7253
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3732344676526211e-06,
+      "loss": 0.6301,
+      "step": 7254
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3713464688644006e-06,
+      "loss": 0.5509,
+      "step": 7255
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.369459673284963e-06,
+      "loss": 0.8985,
+      "step": 7256
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3675740811774118e-06,
+      "loss": 0.7557,
+      "step": 7257
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.365689692804676e-06,
+      "loss": 0.5023,
+      "step": 7258
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3638065084295237e-06,
+      "loss": 0.5677,
+      "step": 7259
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.361924528314551e-06,
+      "loss": 0.7512,
+      "step": 7260
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3600437527221921e-06,
+      "loss": 0.7419,
+      "step": 7261
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.358164181914704e-06,
+      "loss": 0.7056,
+      "step": 7262
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3562858161541804e-06,
+      "loss": 0.5327,
+      "step": 7263
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3544086557025493e-06,
+      "loss": 0.5614,
+      "step": 7264
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3525327008215694e-06,
+      "loss": 0.6991,
+      "step": 7265
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3506579517728247e-06,
+      "loss": 0.6871,
+      "step": 7266
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3487844088177416e-06,
+      "loss": 0.5068,
+      "step": 7267
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3469120722175754e-06,
+      "loss": 0.5502,
+      "step": 7268
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3450409422334043e-06,
+      "loss": 0.6467,
+      "step": 7269
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3431710191261483e-06,
+      "loss": 0.5552,
+      "step": 7270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3413023031565564e-06,
+      "loss": 0.4694,
+      "step": 7271
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3394347945852105e-06,
+      "loss": 0.4951,
+      "step": 7272
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3375684936725175e-06,
+      "loss": 0.5906,
+      "step": 7273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.335703400678724e-06,
+      "loss": 0.6556,
+      "step": 7274
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3338395158639039e-06,
+      "loss": 0.5353,
+      "step": 7275
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3319768394879652e-06,
+      "loss": 0.6366,
+      "step": 7276
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.330115371810643e-06,
+      "loss": 0.5172,
+      "step": 7277
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.328255113091509e-06,
+      "loss": 0.8238,
+      "step": 7278
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.326396063589963e-06,
+      "loss": 0.9003,
+      "step": 7279
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3245382235652393e-06,
+      "loss": 0.5053,
+      "step": 7280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.322681593276397e-06,
+      "loss": 0.5488,
+      "step": 7281
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3208261729823335e-06,
+      "loss": 0.7683,
+      "step": 7282
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3189719629417752e-06,
+      "loss": 0.6751,
+      "step": 7283
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.317118963413282e-06,
+      "loss": 0.5829,
+      "step": 7284
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3152671746552347e-06,
+      "loss": 0.535,
+      "step": 7285
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3134165969258594e-06,
+      "loss": 0.6404,
+      "step": 7286
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3115672304832094e-06,
+      "loss": 0.7483,
+      "step": 7287
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3097190755851596e-06,
+      "loss": 0.6904,
+      "step": 7288
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3078721324894261e-06,
+      "loss": 0.5812,
+      "step": 7289
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.306026401453553e-06,
+      "loss": 0.6689,
+      "step": 7290
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3041818827349172e-06,
+      "loss": 0.5743,
+      "step": 7291
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3023385765907203e-06,
+      "loss": 0.5808,
+      "step": 7292
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.300496483278002e-06,
+      "loss": 0.5431,
+      "step": 7293
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.29865560305363e-06,
+      "loss": 0.6156,
+      "step": 7294
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.296815936174305e-06,
+      "loss": 0.7405,
+      "step": 7295
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.294977482896551e-06,
+      "loss": 0.6079,
+      "step": 7296
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2931402434767315e-06,
+      "loss": 0.6205,
+      "step": 7297
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2913042181710368e-06,
+      "loss": 0.649,
+      "step": 7298
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2894694072354919e-06,
+      "loss": 0.5974,
+      "step": 7299
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2876358109259424e-06,
+      "loss": 0.598,
+      "step": 7300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.285803429498076e-06,
+      "loss": 0.5119,
+      "step": 7301
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2839722632074036e-06,
+      "loss": 0.5794,
+      "step": 7302
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2821423123092714e-06,
+      "loss": 0.8161,
+      "step": 7303
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2803135770588526e-06,
+      "loss": 0.7361,
+      "step": 7304
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2784860577111535e-06,
+      "loss": 0.6958,
+      "step": 7305
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.276659754521009e-06,
+      "loss": 0.6284,
+      "step": 7306
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2748346677430867e-06,
+      "loss": 0.4947,
+      "step": 7307
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2730107976318796e-06,
+      "loss": 0.818,
+      "step": 7308
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2711881444417152e-06,
+      "loss": 0.6002,
+      "step": 7309
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2693667084267513e-06,
+      "loss": 0.6618,
+      "step": 7310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2675464898409772e-06,
+      "loss": 0.5816,
+      "step": 7311
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2657274889382065e-06,
+      "loss": 0.5256,
+      "step": 7312
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2639097059720883e-06,
+      "loss": 0.6076,
+      "step": 7313
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2620931411961002e-06,
+      "loss": 0.6574,
+      "step": 7314
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2602777948635536e-06,
+      "loss": 0.5453,
+      "step": 7315
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2584636672275807e-06,
+      "loss": 0.5422,
+      "step": 7316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2566507585411525e-06,
+      "loss": 0.93,
+      "step": 7317
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2548390690570678e-06,
+      "loss": 0.754,
+      "step": 7318
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2530285990279567e-06,
+      "loss": 0.4918,
+      "step": 7319
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2512193487062718e-06,
+      "loss": 0.5181,
+      "step": 7320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2494113183443013e-06,
+      "loss": 0.6083,
+      "step": 7321
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2476045081941713e-06,
+      "loss": 0.5927,
+      "step": 7322
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2457989185078211e-06,
+      "loss": 0.543,
+      "step": 7323
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2439945495370321e-06,
+      "loss": 0.6387,
+      "step": 7324
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2421914015334103e-06,
+      "loss": 0.8426,
+      "step": 7325
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2403894747483957e-06,
+      "loss": 0.8118,
+      "step": 7326
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2385887694332498e-06,
+      "loss": 0.5853,
+      "step": 7327
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.236789285839073e-06,
+      "loss": 0.5532,
+      "step": 7328
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2349910242167895e-06,
+      "loss": 0.5267,
+      "step": 7329
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2331939848171582e-06,
+      "loss": 0.7608,
+      "step": 7330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2313981678907593e-06,
+      "loss": 0.6856,
+      "step": 7331
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2296035736880118e-06,
+      "loss": 0.616,
+      "step": 7332
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2278102024591576e-06,
+      "loss": 0.5783,
+      "step": 7333
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2260180544542744e-06,
+      "loss": 0.7199,
+      "step": 7334
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2242271299232599e-06,
+      "loss": 0.5126,
+      "step": 7335
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2224374291158492e-06,
+      "loss": 0.498,
+      "step": 7336
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2206489522816045e-06,
+      "loss": 0.585,
+      "step": 7337
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2188616996699187e-06,
+      "loss": 0.6371,
+      "step": 7338
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2170756715300092e-06,
+      "loss": 0.5408,
+      "step": 7339
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2152908681109244e-06,
+      "loss": 0.6063,
+      "step": 7340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2135072896615507e-06,
+      "loss": 0.5706,
+      "step": 7341
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2117249364305905e-06,
+      "loss": 0.4767,
+      "step": 7342
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2099438086665816e-06,
+      "loss": 0.6382,
+      "step": 7343
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2081639066178918e-06,
+      "loss": 0.4945,
+      "step": 7344
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2063852305327183e-06,
+      "loss": 0.8638,
+      "step": 7345
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2046077806590818e-06,
+      "loss": 0.554,
+      "step": 7346
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.202831557244838e-06,
+      "loss": 0.7534,
+      "step": 7347
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.201056560537669e-06,
+      "loss": 0.8056,
+      "step": 7348
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1992827907850902e-06,
+      "loss": 0.5359,
+      "step": 7349
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.197510248234436e-06,
+      "loss": 0.6248,
+      "step": 7350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1957389331328784e-06,
+      "loss": 0.903,
+      "step": 7351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1939688457274157e-06,
+      "loss": 0.6741,
+      "step": 7352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1921999862648769e-06,
+      "loss": 0.5204,
+      "step": 7353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1904323549919138e-06,
+      "loss": 0.546,
+      "step": 7354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1886659521550126e-06,
+      "loss": 0.6725,
+      "step": 7355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1869007780004871e-06,
+      "loss": 0.8993,
+      "step": 7356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1851368327744806e-06,
+      "loss": 0.5183,
+      "step": 7357
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1833741167229584e-06,
+      "loss": 0.8555,
+      "step": 7358
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1816126300917253e-06,
+      "loss": 0.6464,
+      "step": 7359
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1798523731264079e-06,
+      "loss": 0.5871,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.178093346072461e-06,
+      "loss": 0.5386,
+      "step": 7361
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1763355491751683e-06,
+      "loss": 0.7012,
+      "step": 7362
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.174578982679644e-06,
+      "loss": 0.5644,
+      "step": 7363
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1728236468308328e-06,
+      "loss": 0.5753,
+      "step": 7364
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1710695418734996e-06,
+      "loss": 0.4997,
+      "step": 7365
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1693166680522461e-06,
+      "loss": 0.6262,
+      "step": 7366
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1675650256114967e-06,
+      "loss": 0.6458,
+      "step": 7367
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1658146147955096e-06,
+      "loss": 0.6876,
+      "step": 7368
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.164065435848365e-06,
+      "loss": 0.5415,
+      "step": 7369
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1623174890139755e-06,
+      "loss": 0.5407,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1605707745360796e-06,
+      "loss": 0.4216,
+      "step": 7371
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1588252926582489e-06,
+      "loss": 0.5359,
+      "step": 7372
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1570810436238743e-06,
+      "loss": 0.6491,
+      "step": 7373
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1553380276761828e-06,
+      "loss": 0.7736,
+      "step": 7374
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.153596245058225e-06,
+      "loss": 0.6679,
+      "step": 7375
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1518556960128846e-06,
+      "loss": 0.8939,
+      "step": 7376
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.150116380782863e-06,
+      "loss": 0.6486,
+      "step": 7377
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1483782996107006e-06,
+      "loss": 0.7966,
+      "step": 7378
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1466414527387647e-06,
+      "loss": 0.8834,
+      "step": 7379
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1449058404092406e-06,
+      "loss": 0.7557,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.14317146286415e-06,
+      "loss": 0.5063,
+      "step": 7381
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1414383203453416e-06,
+      "loss": 0.6224,
+      "step": 7382
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1397064130944924e-06,
+      "loss": 0.4649,
+      "step": 7383
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1379757413531e-06,
+      "loss": 0.5186,
+      "step": 7384
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1362463053624996e-06,
+      "loss": 0.5142,
+      "step": 7385
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1345181053638488e-06,
+      "loss": 0.6153,
+      "step": 7386
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1327911415981341e-06,
+      "loss": 0.7863,
+      "step": 7387
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1310654143061662e-06,
+      "loss": 0.5111,
+      "step": 7388
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.129340923728589e-06,
+      "loss": 0.4706,
+      "step": 7389
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1276176701058717e-06,
+      "loss": 0.7329,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1258956536783107e-06,
+      "loss": 0.7697,
+      "step": 7391
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1241748746860271e-06,
+      "loss": 0.8168,
+      "step": 7392
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1224553333689747e-06,
+      "loss": 0.6337,
+      "step": 7393
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1207370299669308e-06,
+      "loss": 0.5132,
+      "step": 7394
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1190199647195022e-06,
+      "loss": 0.8169,
+      "step": 7395
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1173041378661231e-06,
+      "loss": 0.6284,
+      "step": 7396
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1155895496460522e-06,
+      "loss": 0.5572,
+      "step": 7397
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1138762002983816e-06,
+      "loss": 0.8489,
+      "step": 7398
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1121640900620211e-06,
+      "loss": 0.8007,
+      "step": 7399
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1104532191757155e-06,
+      "loss": 0.8488,
+      "step": 7400
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1087435878780362e-06,
+      "loss": 0.5524,
+      "step": 7401
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.107035196407379e-06,
+      "loss": 0.5628,
+      "step": 7402
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1053280450019653e-06,
+      "loss": 0.6174,
+      "step": 7403
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1036221338998487e-06,
+      "loss": 0.7773,
+      "step": 7404
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1019174633389073e-06,
+      "loss": 0.5909,
+      "step": 7405
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1002140335568467e-06,
+      "loss": 0.7077,
+      "step": 7406
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0985118447911958e-06,
+      "loss": 0.5517,
+      "step": 7407
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0968108972793157e-06,
+      "loss": 0.5477,
+      "step": 7408
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0951111912583922e-06,
+      "loss": 0.868,
+      "step": 7409
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0934127269654392e-06,
+      "loss": 0.7485,
+      "step": 7410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.091715504637294e-06,
+      "loss": 0.6049,
+      "step": 7411
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.090019524510625e-06,
+      "loss": 0.7408,
+      "step": 7412
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0883247868219226e-06,
+      "loss": 0.5526,
+      "step": 7413
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0866312918075106e-06,
+      "loss": 0.5241,
+      "step": 7414
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0849390397035331e-06,
+      "loss": 0.5672,
+      "step": 7415
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0832480307459648e-06,
+      "loss": 0.5405,
+      "step": 7416
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.081558265170607e-06,
+      "loss": 0.5176,
+      "step": 7417
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0798697432130822e-06,
+      "loss": 0.6303,
+      "step": 7418
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0781824651088468e-06,
+      "loss": 0.4596,
+      "step": 7419
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0764964310931792e-06,
+      "loss": 0.5322,
+      "step": 7420
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.074811641401189e-06,
+      "loss": 0.611,
+      "step": 7421
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0731280962678047e-06,
+      "loss": 0.6111,
+      "step": 7422
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0714457959277857e-06,
+      "loss": 0.793,
+      "step": 7423
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0697647406157209e-06,
+      "loss": 0.525,
+      "step": 7424
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0680849305660213e-06,
+      "loss": 0.7738,
+      "step": 7425
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0664063660129243e-06,
+      "loss": 0.512,
+      "step": 7426
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0647290471904936e-06,
+      "loss": 0.6486,
+      "step": 7427
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0630529743326223e-06,
+      "loss": 0.6966,
+      "step": 7428
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.061378147673029e-06,
+      "loss": 0.5021,
+      "step": 7429
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0597045674452534e-06,
+      "loss": 0.5021,
+      "step": 7430
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0580322338826666e-06,
+      "loss": 0.6138,
+      "step": 7431
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.056361147218463e-06,
+      "loss": 0.4542,
+      "step": 7432
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0546913076856712e-06,
+      "loss": 0.6012,
+      "step": 7433
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0530227155171336e-06,
+      "loss": 0.5658,
+      "step": 7434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0513553709455237e-06,
+      "loss": 0.6613,
+      "step": 7435
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0496892742033448e-06,
+      "loss": 0.704,
+      "step": 7436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0480244255229234e-06,
+      "loss": 0.475,
+      "step": 7437
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0463608251364076e-06,
+      "loss": 0.5501,
+      "step": 7438
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.044698473275778e-06,
+      "loss": 0.7171,
+      "step": 7439
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.043037370172838e-06,
+      "loss": 0.6827,
+      "step": 7440
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.041377516059221e-06,
+      "loss": 0.5748,
+      "step": 7441
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0397189111663775e-06,
+      "loss": 0.5325,
+      "step": 7442
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0380615557255903e-06,
+      "loss": 0.5208,
+      "step": 7443
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0364054499679688e-06,
+      "loss": 0.6075,
+      "step": 7444
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0347505941244462e-06,
+      "loss": 0.6859,
+      "step": 7445
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.033096988425778e-06,
+      "loss": 0.9444,
+      "step": 7446
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0314446331025496e-06,
+      "loss": 0.4967,
+      "step": 7447
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0297935283851722e-06,
+      "loss": 0.4615,
+      "step": 7448
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0281436745038842e-06,
+      "loss": 0.8258,
+      "step": 7449
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.026495071688739e-06,
+      "loss": 0.4917,
+      "step": 7450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0248477201696294e-06,
+      "loss": 0.538,
+      "step": 7451
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0232016201762696e-06,
+      "loss": 0.5678,
+      "step": 7452
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0215567719381914e-06,
+      "loss": 0.6634,
+      "step": 7453
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0199131756847613e-06,
+      "loss": 0.5467,
+      "step": 7454
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0182708316451672e-06,
+      "loss": 0.6053,
+      "step": 7455
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.016629740048426e-06,
+      "loss": 0.8301,
+      "step": 7456
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0149899011233711e-06,
+      "loss": 0.8563,
+      "step": 7457
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0133513150986717e-06,
+      "loss": 0.5938,
+      "step": 7458
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0117139822028166e-06,
+      "loss": 0.8492,
+      "step": 7459
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0100779026641227e-06,
+      "loss": 0.5078,
+      "step": 7460
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.008443076710728e-06,
+      "loss": 0.4886,
+      "step": 7461
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0068095045705982e-06,
+      "loss": 0.5324,
+      "step": 7462
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0051771864715254e-06,
+      "loss": 0.7272,
+      "step": 7463
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0035461226411282e-06,
+      "loss": 0.5364,
+      "step": 7464
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.001916313306841e-06,
+      "loss": 0.6242,
+      "step": 7465
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0002877586959348e-06,
+      "loss": 0.6374,
+      "step": 7466
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.986604590354999e-07,
+      "loss": 0.7833,
+      "step": 7467
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.970344145524546e-07,
+      "loss": 0.5107,
+      "step": 7468
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.954096254735335e-07,
+      "loss": 0.5036,
+      "step": 7469
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.937860920253084e-07,
+      "loss": 0.873,
+      "step": 7470
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.921638144341715e-07,
+      "loss": 0.5583,
+      "step": 7471
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.905427929263333e-07,
+      "loss": 0.5499,
+      "step": 7472
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.889230277278384e-07,
+      "loss": 0.5714,
+      "step": 7473
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.873045190645514e-07,
+      "loss": 0.9741,
+      "step": 7474
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.856872671621643e-07,
+      "loss": 0.5514,
+      "step": 7475
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.840712722461886e-07,
+      "loss": 0.6587,
+      "step": 7476
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.824565345419656e-07,
+      "loss": 0.6623,
+      "step": 7477
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.808430542746605e-07,
+      "loss": 0.7928,
+      "step": 7478
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.792308316692645e-07,
+      "loss": 0.5424,
+      "step": 7479
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.776198669505876e-07,
+      "loss": 0.6942,
+      "step": 7480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.760101603432682e-07,
+      "loss": 0.5575,
+      "step": 7481
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.744017120717719e-07,
+      "loss": 0.5432,
+      "step": 7482
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.727945223603874e-07,
+      "loss": 0.5082,
+      "step": 7483
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.71188591433222e-07,
+      "loss": 0.5612,
+      "step": 7484
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.695839195142142e-07,
+      "loss": 0.5395,
+      "step": 7485
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.679805068271252e-07,
+      "loss": 0.6008,
+      "step": 7486
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.663783535955407e-07,
+      "loss": 0.524,
+      "step": 7487
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.647774600428705e-07,
+      "loss": 0.7411,
+      "step": 7488
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.631778263923475e-07,
+      "loss": 0.6187,
+      "step": 7489
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.615794528670331e-07,
+      "loss": 0.8547,
+      "step": 7490
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.599823396898055e-07,
+      "loss": 0.6083,
+      "step": 7491
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.583864870833737e-07,
+      "loss": 0.6723,
+      "step": 7492
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.567918952702681e-07,
+      "loss": 0.5776,
+      "step": 7493
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.55198564472848e-07,
+      "loss": 0.5328,
+      "step": 7494
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.536064949132862e-07,
+      "loss": 0.6366,
+      "step": 7495
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.52015686813591e-07,
+      "loss": 0.8967,
+      "step": 7496
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.504261403955883e-07,
+      "loss": 0.5445,
+      "step": 7497
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.488378558809319e-07,
+      "loss": 0.5268,
+      "step": 7498
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.472508334910946e-07,
+      "loss": 0.7715,
+      "step": 7499
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.456650734473771e-07,
+      "loss": 0.5359,
+      "step": 7500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.440805759709037e-07,
+      "loss": 0.6624,
+      "step": 7501
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.424973412826244e-07,
+      "loss": 0.5173,
+      "step": 7502
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.409153696033058e-07,
+      "loss": 0.6375,
+      "step": 7503
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.39334661153547e-07,
+      "loss": 0.54,
+      "step": 7504
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.377552161537662e-07,
+      "loss": 0.5603,
+      "step": 7505
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.361770348242072e-07,
+      "loss": 0.5194,
+      "step": 7506
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.346001173849362e-07,
+      "loss": 0.5495,
+      "step": 7507
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.33024464055844e-07,
+      "loss": 0.5498,
+      "step": 7508
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.314500750566479e-07,
+      "loss": 0.7519,
+      "step": 7509
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.298769506068816e-07,
+      "loss": 0.8318,
+      "step": 7510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.283050909259095e-07,
+      "loss": 0.5777,
+      "step": 7511
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.26734496232915e-07,
+      "loss": 0.7014,
+      "step": 7512
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.25165166746912e-07,
+      "loss": 0.4983,
+      "step": 7513
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.235971026867264e-07,
+      "loss": 0.5099,
+      "step": 7514
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.220303042710188e-07,
+      "loss": 0.462,
+      "step": 7515
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.20464771718268e-07,
+      "loss": 0.5558,
+      "step": 7516
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.189005052467781e-07,
+      "loss": 0.7552,
+      "step": 7517
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.173375050746725e-07,
+      "loss": 0.6085,
+      "step": 7518
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.157757714199045e-07,
+      "loss": 0.6859,
+      "step": 7519
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.142153045002456e-07,
+      "loss": 0.5313,
+      "step": 7520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.12656104533296e-07,
+      "loss": 0.5659,
+      "step": 7521
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.110981717364709e-07,
+      "loss": 0.6216,
+      "step": 7522
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.095415063270163e-07,
+      "loss": 0.6118,
+      "step": 7523
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.079861085219999e-07,
+      "loss": 0.9103,
+      "step": 7524
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.064319785383091e-07,
+      "loss": 0.7976,
+      "step": 7525
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.048791165926584e-07,
+      "loss": 0.6259,
+      "step": 7526
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.033275229015847e-07,
+      "loss": 0.5538,
+      "step": 7527
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.017771976814493e-07,
+      "loss": 0.6029,
+      "step": 7528
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.002281411484303e-07,
+      "loss": 0.6112,
+      "step": 7529
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.986803535185363e-07,
+      "loss": 0.5189,
+      "step": 7530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.971338350075953e-07,
+      "loss": 0.6805,
+      "step": 7531
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.955885858312607e-07,
+      "loss": 0.493,
+      "step": 7532
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.940446062050034e-07,
+      "loss": 0.624,
+      "step": 7533
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.925018963441245e-07,
+      "loss": 0.5668,
+      "step": 7534
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.909604564637442e-07,
+      "loss": 0.58,
+      "step": 7535
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.894202867788059e-07,
+      "loss": 0.8836,
+      "step": 7536
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.878813875040759e-07,
+      "loss": 0.4761,
+      "step": 7537
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.863437588541423e-07,
+      "loss": 0.6042,
+      "step": 7538
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.848074010434194e-07,
+      "loss": 0.5536,
+      "step": 7539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.832723142861432e-07,
+      "loss": 0.6814,
+      "step": 7540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.817384987963673e-07,
+      "loss": 0.5435,
+      "step": 7541
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.802059547879727e-07,
+      "loss": 0.6473,
+      "step": 7542
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.786746824746662e-07,
+      "loss": 0.8717,
+      "step": 7543
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.771446820699725e-07,
+      "loss": 0.5989,
+      "step": 7544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.756159537872389e-07,
+      "loss": 0.4666,
+      "step": 7545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.740884978396358e-07,
+      "loss": 0.6903,
+      "step": 7546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.725623144401574e-07,
+      "loss": 0.5217,
+      "step": 7547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.710374038016223e-07,
+      "loss": 0.6645,
+      "step": 7548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.69513766136666e-07,
+      "loss": 0.493,
+      "step": 7549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.679914016577507e-07,
+      "loss": 0.5701,
+      "step": 7550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.66470310577161e-07,
+      "loss": 0.4263,
+      "step": 7551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.64950493107003e-07,
+      "loss": 0.5481,
+      "step": 7552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.634319494592036e-07,
+      "loss": 0.6021,
+      "step": 7553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.619146798455147e-07,
+      "loss": 0.6677,
+      "step": 7554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.603986844775092e-07,
+      "loss": 0.7778,
+      "step": 7555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.588839635665847e-07,
+      "loss": 0.6151,
+      "step": 7556
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.573705173239543e-07,
+      "loss": 0.6473,
+      "step": 7557
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.558583459606629e-07,
+      "loss": 0.5232,
+      "step": 7558
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.543474496875703e-07,
+      "loss": 0.648,
+      "step": 7559
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.528378287153626e-07,
+      "loss": 0.4979,
+      "step": 7560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.513294832545438e-07,
+      "loss": 0.5685,
+      "step": 7561
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.498224135154454e-07,
+      "loss": 0.4226,
+      "step": 7562
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.483166197082183e-07,
+      "loss": 0.6194,
+      "step": 7563
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.468121020428332e-07,
+      "loss": 0.5504,
+      "step": 7564
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.453088607290871e-07,
+      "loss": 0.6047,
+      "step": 7565
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.438068959765966e-07,
+      "loss": 0.6195,
+      "step": 7566
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.42306207994803e-07,
+      "loss": 0.6245,
+      "step": 7567
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.408067969929623e-07,
+      "loss": 0.5795,
+      "step": 7568
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.393086631801606e-07,
+      "loss": 0.5828,
+      "step": 7569
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.378118067653018e-07,
+      "loss": 0.5523,
+      "step": 7570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.363162279571157e-07,
+      "loss": 0.5616,
+      "step": 7571
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.348219269641466e-07,
+      "loss": 0.6667,
+      "step": 7572
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.333289039947668e-07,
+      "loss": 0.4803,
+      "step": 7573
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.318371592571683e-07,
+      "loss": 0.8382,
+      "step": 7574
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.303466929593662e-07,
+      "loss": 0.8149,
+      "step": 7575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.28857505309194e-07,
+      "loss": 0.603,
+      "step": 7576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.273695965143114e-07,
+      "loss": 0.6112,
+      "step": 7577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.258829667821955e-07,
+      "loss": 0.5314,
+      "step": 7578
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.243976163201473e-07,
+      "loss": 0.806,
+      "step": 7579
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.229135453352898e-07,
+      "loss": 0.6307,
+      "step": 7580
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.214307540345679e-07,
+      "loss": 0.5609,
+      "step": 7581
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.199492426247468e-07,
+      "loss": 0.4824,
+      "step": 7582
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.184690113124105e-07,
+      "loss": 0.7461,
+      "step": 7583
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.169900603039704e-07,
+      "loss": 0.6221,
+      "step": 7584
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.15512389805655e-07,
+      "loss": 0.5098,
+      "step": 7585
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.140360000235181e-07,
+      "loss": 0.7758,
+      "step": 7586
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.125608911634297e-07,
+      "loss": 0.5838,
+      "step": 7587
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.110870634310841e-07,
+      "loss": 0.8299,
+      "step": 7588
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.096145170319991e-07,
+      "loss": 0.7749,
+      "step": 7589
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.081432521715104e-07,
+      "loss": 0.6851,
+      "step": 7590
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.066732690547763e-07,
+      "loss": 0.5924,
+      "step": 7591
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.052045678867748e-07,
+      "loss": 0.6716,
+      "step": 7592
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.037371488723078e-07,
+      "loss": 0.5478,
+      "step": 7593
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.022710122159993e-07,
+      "loss": 0.5741,
+      "step": 7594
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.008061581222881e-07,
+      "loss": 0.5929,
+      "step": 7595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.993425867954407e-07,
+      "loss": 0.5095,
+      "step": 7596
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.978802984395417e-07,
+      "loss": 0.703,
+      "step": 7597
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.964192932584991e-07,
+      "loss": 0.6059,
+      "step": 7598
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.949595714560399e-07,
+      "loss": 0.935,
+      "step": 7599
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.935011332357113e-07,
+      "loss": 0.6854,
+      "step": 7600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.920439788008871e-07,
+      "loss": 0.5034,
+      "step": 7601
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.905881083547518e-07,
+      "loss": 0.5746,
+      "step": 7602
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.891335221003193e-07,
+      "loss": 0.5934,
+      "step": 7603
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.876802202404232e-07,
+      "loss": 0.6441,
+      "step": 7604
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.86228202977718e-07,
+      "loss": 0.475,
+      "step": 7605
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.847774705146749e-07,
+      "loss": 0.7056,
+      "step": 7606
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8332802305359e-07,
+      "loss": 0.841,
+      "step": 7607
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818798607965782e-07,
+      "loss": 0.6242,
+      "step": 7608
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804329839455804e-07,
+      "loss": 0.6192,
+      "step": 7609
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.789873927023484e-07,
+      "loss": 0.4602,
+      "step": 7610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.775430872684631e-07,
+      "loss": 0.6419,
+      "step": 7611
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.761000678453234e-07,
+      "loss": 0.5894,
+      "step": 7612
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.746583346341507e-07,
+      "loss": 0.9179,
+      "step": 7613
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732178878359808e-07,
+      "loss": 0.6751,
+      "step": 7614
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.717787276516775e-07,
+      "loss": 0.5539,
+      "step": 7615
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703408542819202e-07,
+      "loss": 0.5215,
+      "step": 7616
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.68904267927213e-07,
+      "loss": 0.6014,
+      "step": 7617
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674689687878778e-07,
+      "loss": 0.4715,
+      "step": 7618
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660349570640569e-07,
+      "loss": 0.4955,
+      "step": 7619
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.646022329557156e-07,
+      "loss": 0.5761,
+      "step": 7620
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.631707966626356e-07,
+      "loss": 0.4426,
+      "step": 7621
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.617406483844214e-07,
+      "loss": 0.784,
+      "step": 7622
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.60311788320498e-07,
+      "loss": 0.6279,
+      "step": 7623
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.588842166701138e-07,
+      "loss": 0.5951,
+      "step": 7624
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.574579336323284e-07,
+      "loss": 0.5057,
+      "step": 7625
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.560329394060317e-07,
+      "loss": 0.7634,
+      "step": 7626
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.546092341899269e-07,
+      "loss": 0.5634,
+      "step": 7627
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.531868181825452e-07,
+      "loss": 0.6111,
+      "step": 7628
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.517656915822269e-07,
+      "loss": 0.6298,
+      "step": 7629
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.503458545871423e-07,
+      "loss": 0.5818,
+      "step": 7630
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.489273073952785e-07,
+      "loss": 0.8845,
+      "step": 7631
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.475100502044419e-07,
+      "loss": 0.6983,
+      "step": 7632
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.460940832122587e-07,
+      "loss": 0.7696,
+      "step": 7633
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.446794066161756e-07,
+      "loss": 0.8344,
+      "step": 7634
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.432660206134634e-07,
+      "loss": 0.6781,
+      "step": 7635
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.418539254012058e-07,
+      "loss": 0.7988,
+      "step": 7636
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.404431211763131e-07,
+      "loss": 0.66,
+      "step": 7637
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.390336081355098e-07,
+      "loss": 0.579,
+      "step": 7638
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.376253864753469e-07,
+      "loss": 0.6861,
+      "step": 7639
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.36218456392187e-07,
+      "loss": 0.8926,
+      "step": 7640
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.348128180822201e-07,
+      "loss": 0.5107,
+      "step": 7641
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.334084717414525e-07,
+      "loss": 0.9578,
+      "step": 7642
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.320054175657121e-07,
+      "loss": 0.5395,
+      "step": 7643
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.306036557506435e-07,
+      "loss": 0.7878,
+      "step": 7644
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.292031864917137e-07,
+      "loss": 0.5534,
+      "step": 7645
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.278040099842099e-07,
+      "loss": 0.626,
+      "step": 7646
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.264061264232391e-07,
+      "loss": 0.4812,
+      "step": 7647
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.250095360037235e-07,
+      "loss": 0.5024,
+      "step": 7648
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.236142389204104e-07,
+      "loss": 0.8818,
+      "step": 7649
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.222202353678653e-07,
+      "loss": 0.7491,
+      "step": 7650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.208275255404728e-07,
+      "loss": 0.6983,
+      "step": 7651
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.194361096324354e-07,
+      "loss": 0.6523,
+      "step": 7652
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.180459878377754e-07,
+      "loss": 0.5584,
+      "step": 7653
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.166571603503414e-07,
+      "loss": 0.6148,
+      "step": 7654
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.152696273637949e-07,
+      "loss": 0.5717,
+      "step": 7655
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.138833890716146e-07,
+      "loss": 0.5391,
+      "step": 7656
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.124984456671047e-07,
+      "loss": 0.5728,
+      "step": 7657
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.111147973433852e-07,
+      "loss": 0.5979,
+      "step": 7658
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.097324442934006e-07,
+      "loss": 0.509,
+      "step": 7659
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.083513867099056e-07,
+      "loss": 0.8584,
+      "step": 7660
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.069716247854819e-07,
+      "loss": 0.8044,
+      "step": 7661
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.055931587125287e-07,
+      "loss": 0.6032,
+      "step": 7662
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.042159886832655e-07,
+      "loss": 0.546,
+      "step": 7663
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.028401148897257e-07,
+      "loss": 0.731,
+      "step": 7664
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.014655375237666e-07,
+      "loss": 0.6621,
+      "step": 7665
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.000922567770663e-07,
+      "loss": 0.5006,
+      "step": 7666
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.987202728411202e-07,
+      "loss": 0.5945,
+      "step": 7667
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.973495859072387e-07,
+      "loss": 0.5031,
+      "step": 7668
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.959801961665569e-07,
+      "loss": 0.5465,
+      "step": 7669
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.946121038100273e-07,
+      "loss": 0.483,
+      "step": 7670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.932453090284208e-07,
+      "loss": 0.7048,
+      "step": 7671
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.918798120123293e-07,
+      "loss": 0.5708,
+      "step": 7672
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.905156129521618e-07,
+      "loss": 0.5147,
+      "step": 7673
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.89152712038148e-07,
+      "loss": 0.5629,
+      "step": 7674
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.877911094603317e-07,
+      "loss": 0.6063,
+      "step": 7675
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.864308054085822e-07,
+      "loss": 0.4926,
+      "step": 7676
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.850718000725843e-07,
+      "loss": 0.758,
+      "step": 7677
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.837140936418441e-07,
+      "loss": 0.5635,
+      "step": 7678
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.823576863056813e-07,
+      "loss": 0.6353,
+      "step": 7679
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.810025782532404e-07,
+      "loss": 0.447,
+      "step": 7680
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.79648769673481e-07,
+      "loss": 0.5603,
+      "step": 7681
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.782962607551857e-07,
+      "loss": 0.7345,
+      "step": 7682
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.769450516869491e-07,
+      "loss": 0.5231,
+      "step": 7683
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.755951426571894e-07,
+      "loss": 0.5298,
+      "step": 7684
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.742465338541426e-07,
+      "loss": 0.5485,
+      "step": 7685
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.728992254658662e-07,
+      "loss": 0.593,
+      "step": 7686
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.715532176802298e-07,
+      "loss": 0.5584,
+      "step": 7687
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.702085106849255e-07,
+      "loss": 0.6624,
+      "step": 7688
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.688651046674655e-07,
+      "loss": 0.6504,
+      "step": 7689
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.675229998151778e-07,
+      "loss": 0.567,
+      "step": 7690
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.661821963152115e-07,
+      "loss": 0.6919,
+      "step": 7691
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.648426943545306e-07,
+      "loss": 0.5501,
+      "step": 7692
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.635044941199231e-07,
+      "loss": 0.6849,
+      "step": 7693
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.62167595797989e-07,
+      "loss": 0.7049,
+      "step": 7694
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.608319995751511e-07,
+      "loss": 0.5374,
+      "step": 7695
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.594977056376483e-07,
+      "loss": 0.6307,
+      "step": 7696
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.58164714171543e-07,
+      "loss": 0.6593,
+      "step": 7697
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.568330253627064e-07,
+      "loss": 0.6668,
+      "step": 7698
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.555026393968377e-07,
+      "loss": 0.5871,
+      "step": 7699
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.541735564594487e-07,
+      "loss": 0.7658,
+      "step": 7700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.528457767358742e-07,
+      "loss": 0.8045,
+      "step": 7701
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.515193004112596e-07,
+      "loss": 0.5995,
+      "step": 7702
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.501941276705759e-07,
+      "loss": 0.564,
+      "step": 7703
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.488702586986107e-07,
+      "loss": 0.6954,
+      "step": 7704
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.475476936799674e-07,
+      "loss": 0.6953,
+      "step": 7705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.462264327990686e-07,
+      "loss": 0.5959,
+      "step": 7706
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.449064762401546e-07,
+      "loss": 0.5963,
+      "step": 7707
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.435878241872873e-07,
+      "loss": 0.5354,
+      "step": 7708
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.422704768243427e-07,
+      "loss": 0.8963,
+      "step": 7709
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.409544343350149e-07,
+      "loss": 0.5755,
+      "step": 7710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.396396969028196e-07,
+      "loss": 0.7535,
+      "step": 7711
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.383262647110888e-07,
+      "loss": 0.6033,
+      "step": 7712
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.370141379429684e-07,
+      "loss": 0.4594,
+      "step": 7713
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.357033167814286e-07,
+      "loss": 0.5497,
+      "step": 7714
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.343938014092533e-07,
+      "loss": 0.6945,
+      "step": 7715
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.330855920090484e-07,
+      "loss": 0.5572,
+      "step": 7716
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.317786887632315e-07,
+      "loss": 0.806,
+      "step": 7717
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.304730918540425e-07,
+      "loss": 0.5877,
+      "step": 7718
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.2916880146354e-07,
+      "loss": 0.5228,
+      "step": 7719
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.278658177735997e-07,
+      "loss": 0.6229,
+      "step": 7720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.265641409659096e-07,
+      "loss": 0.6305,
+      "step": 7721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.252637712219822e-07,
+      "loss": 0.5593,
+      "step": 7722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.23964708723146e-07,
+      "loss": 0.5877,
+      "step": 7723
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.226669536505492e-07,
+      "loss": 0.7127,
+      "step": 7724
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.213705061851494e-07,
+      "loss": 0.6002,
+      "step": 7725
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.200753665077297e-07,
+      "loss": 0.5771,
+      "step": 7726
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.187815347988924e-07,
+      "loss": 0.5458,
+      "step": 7727
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.174890112390497e-07,
+      "loss": 0.5615,
+      "step": 7728
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.161977960084364e-07,
+      "loss": 0.7675,
+      "step": 7729
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.149078892871052e-07,
+      "loss": 0.5123,
+      "step": 7730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.136192912549255e-07,
+      "loss": 0.8489,
+      "step": 7731
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.123320020915813e-07,
+      "loss": 0.4908,
+      "step": 7732
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.110460219765779e-07,
+      "loss": 0.5314,
+      "step": 7733
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.097613510892364e-07,
+      "loss": 0.648,
+      "step": 7734
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.084779896086979e-07,
+      "loss": 0.816,
+      "step": 7735
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.071959377139147e-07,
+      "loss": 0.6788,
+      "step": 7736
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.059151955836628e-07,
+      "loss": 0.6774,
+      "step": 7737
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.046357633965328e-07,
+      "loss": 0.5234,
+      "step": 7738
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.03357641330935e-07,
+      "loss": 0.5357,
+      "step": 7739
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.020808295650926e-07,
+      "loss": 0.5609,
+      "step": 7740
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.008053282770476e-07,
+      "loss": 0.6046,
+      "step": 7741
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.995311376446633e-07,
+      "loss": 0.6642,
+      "step": 7742
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.982582578456175e-07,
+      "loss": 0.5043,
+      "step": 7743
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.969866890574005e-07,
+      "loss": 0.6641,
+      "step": 7744
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.957164314573261e-07,
+      "loss": 0.9503,
+      "step": 7745
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.944474852225268e-07,
+      "loss": 0.9325,
+      "step": 7746
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.931798505299458e-07,
+      "loss": 0.568,
+      "step": 7747
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.919135275563448e-07,
+      "loss": 0.5315,
+      "step": 7748
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.906485164783071e-07,
+      "loss": 0.5854,
+      "step": 7749
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.893848174722305e-07,
+      "loss": 0.7706,
+      "step": 7750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.881224307143274e-07,
+      "loss": 0.6078,
+      "step": 7751
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.868613563806291e-07,
+      "loss": 0.5976,
+      "step": 7752
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.856015946469851e-07,
+      "loss": 0.5229,
+      "step": 7753
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.843431456890624e-07,
+      "loss": 0.6337,
+      "step": 7754
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.830860096823398e-07,
+      "loss": 0.778,
+      "step": 7755
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.818301868021192e-07,
+      "loss": 0.6639,
+      "step": 7756
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.805756772235161e-07,
+      "loss": 0.8841,
+      "step": 7757
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.793224811214649e-07,
+      "loss": 0.8172,
+      "step": 7758
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.780705986707124e-07,
+      "loss": 0.6295,
+      "step": 7759
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.768200300458282e-07,
+      "loss": 0.6355,
+      "step": 7760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.755707754211948e-07,
+      "loss": 0.7601,
+      "step": 7761
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.743228349710128e-07,
+      "loss": 0.4925,
+      "step": 7762
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.730762088692976e-07,
+      "loss": 0.8853,
+      "step": 7763
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.718308972898856e-07,
+      "loss": 0.7232,
+      "step": 7764
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.705869004064257e-07,
+      "loss": 0.8803,
+      "step": 7765
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.693442183923869e-07,
+      "loss": 0.5154,
+      "step": 7766
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.681028514210496e-07,
+      "loss": 0.5378,
+      "step": 7767
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.668627996655174e-07,
+      "loss": 0.7971,
+      "step": 7768
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.656240632987053e-07,
+      "loss": 0.6771,
+      "step": 7769
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.643866424933498e-07,
+      "loss": 0.6447,
+      "step": 7770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.63150537421997e-07,
+      "loss": 0.8449,
+      "step": 7771
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.619157482570159e-07,
+      "loss": 0.9434,
+      "step": 7772
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.606822751705899e-07,
+      "loss": 0.5568,
+      "step": 7773
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.594501183347189e-07,
+      "loss": 0.5669,
+      "step": 7774
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.58219277921217e-07,
+      "loss": 0.5698,
+      "step": 7775
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.569897541017178e-07,
+      "loss": 0.7601,
+      "step": 7776
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.557615470476719e-07,
+      "loss": 0.5254,
+      "step": 7777
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.545346569303434e-07,
+      "loss": 0.5916,
+      "step": 7778
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.533090839208133e-07,
+      "loss": 0.4911,
+      "step": 7779
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.520848281899793e-07,
+      "loss": 0.8263,
+      "step": 7780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.508618899085583e-07,
+      "loss": 0.7887,
+      "step": 7781
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.496402692470781e-07,
+      "loss": 0.5625,
+      "step": 7782
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.484199663758883e-07,
+      "loss": 0.851,
+      "step": 7783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.472009814651491e-07,
+      "loss": 0.5415,
+      "step": 7784
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.459833146848447e-07,
+      "loss": 0.5282,
+      "step": 7785
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.44766966204765e-07,
+      "loss": 0.554,
+      "step": 7786
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.435519361945251e-07,
+      "loss": 0.4993,
+      "step": 7787
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.423382248235521e-07,
+      "loss": 0.5883,
+      "step": 7788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.411258322610912e-07,
+      "loss": 0.7374,
+      "step": 7789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.39914758676201e-07,
+      "loss": 0.639,
+      "step": 7790
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.387050042377572e-07,
+      "loss": 0.4659,
+      "step": 7791
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.37496569114454e-07,
+      "loss": 0.4987,
+      "step": 7792
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.362894534747998e-07,
+      "loss": 0.599,
+      "step": 7793
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.350836574871154e-07,
+      "loss": 0.8364,
+      "step": 7794
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.338791813195443e-07,
+      "loss": 0.6216,
+      "step": 7795
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.32676025140042e-07,
+      "loss": 0.5398,
+      "step": 7796
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.314741891163822e-07,
+      "loss": 0.4824,
+      "step": 7797
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.302736734161507e-07,
+      "loss": 0.6275,
+      "step": 7798
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.290744782067514e-07,
+      "loss": 0.5308,
+      "step": 7799
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.278766036554051e-07,
+      "loss": 0.5242,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.26680049929148e-07,
+      "loss": 0.71,
+      "step": 7801
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.254848171948312e-07,
+      "loss": 0.514,
+      "step": 7802
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.242909056191214e-07,
+      "loss": 0.6729,
+      "step": 7803
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.230983153685032e-07,
+      "loss": 0.5366,
+      "step": 7804
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.219070466092735e-07,
+      "loss": 0.5603,
+      "step": 7805
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.207170995075473e-07,
+      "loss": 0.5904,
+      "step": 7806
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.195284742292562e-07,
+      "loss": 0.8551,
+      "step": 7807
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.183411709401465e-07,
+      "loss": 0.6076,
+      "step": 7808
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.171551898057769e-07,
+      "loss": 0.5154,
+      "step": 7809
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.159705309915263e-07,
+      "loss": 0.6051,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.147871946625871e-07,
+      "loss": 0.665,
+      "step": 7811
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.136051809839704e-07,
+      "loss": 0.5852,
+      "step": 7812
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.124244901204966e-07,
+      "loss": 0.635,
+      "step": 7813
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.112451222368064e-07,
+      "loss": 0.5739,
+      "step": 7814
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.100670774973559e-07,
+      "loss": 0.7148,
+      "step": 7815
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.08890356066416e-07,
+      "loss": 0.7939,
+      "step": 7816
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.077149581080698e-07,
+      "loss": 0.5516,
+      "step": 7817
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.065408837862196e-07,
+      "loss": 0.5855,
+      "step": 7818
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.053681332645854e-07,
+      "loss": 0.5181,
+      "step": 7819
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.041967067066966e-07,
+      "loss": 0.5733,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.030266042759024e-07,
+      "loss": 0.5999,
+      "step": 7821
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.018578261353646e-07,
+      "loss": 0.6836,
+      "step": 7822
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.006903724480639e-07,
+      "loss": 0.6008,
+      "step": 7823
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.995242433767911e-07,
+      "loss": 0.7255,
+      "step": 7824
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.983594390841562e-07,
+      "loss": 0.8017,
+      "step": 7825
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.971959597325837e-07,
+      "loss": 0.5216,
+      "step": 7826
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.960338054843139e-07,
+      "loss": 0.6207,
+      "step": 7827
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.948729765014004e-07,
+      "loss": 0.9235,
+      "step": 7828
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.937134729457127e-07,
+      "loss": 0.5397,
+      "step": 7829
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.925552949789381e-07,
+      "loss": 0.5833,
+      "step": 7830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.913984427625751e-07,
+      "loss": 0.7398,
+      "step": 7831
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.902429164579392e-07,
+      "loss": 0.5649,
+      "step": 7832
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.890887162261614e-07,
+      "loss": 0.5574,
+      "step": 7833
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.879358422281866e-07,
+      "loss": 0.7994,
+      "step": 7834
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.867842946247781e-07,
+      "loss": 0.5426,
+      "step": 7835
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.856340735765075e-07,
+      "loss": 0.5417,
+      "step": 7836
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.844851792437666e-07,
+      "loss": 0.7174,
+      "step": 7837
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.833376117867639e-07,
+      "loss": 0.5943,
+      "step": 7838
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.82191371365518e-07,
+      "loss": 0.5211,
+      "step": 7839
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.810464581398644e-07,
+      "loss": 0.5072,
+      "step": 7840
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.799028722694543e-07,
+      "loss": 0.5546,
+      "step": 7841
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.787606139137546e-07,
+      "loss": 0.5241,
+      "step": 7842
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.776196832320424e-07,
+      "loss": 0.5614,
+      "step": 7843
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7648008038341467e-07,
+      "loss": 0.8491,
+      "step": 7844
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.753418055267811e-07,
+      "loss": 0.5689,
+      "step": 7845
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.742048588208692e-07,
+      "loss": 0.5885,
+      "step": 7846
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.730692404242143e-07,
+      "loss": 0.5964,
+      "step": 7847
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.719349504951742e-07,
+      "loss": 0.6031,
+      "step": 7848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.708019891919169e-07,
+      "loss": 0.5758,
+      "step": 7849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.69670356672427e-07,
+      "loss": 0.6844,
+      "step": 7850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.685400530945017e-07,
+      "loss": 0.5599,
+      "step": 7851
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6741107861575597e-07,
+      "loss": 0.4848,
+      "step": 7852
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6628343339361613e-07,
+      "loss": 0.6186,
+      "step": 7853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6515711758532866e-07,
+      "loss": 0.57,
+      "step": 7854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6403213134794457e-07,
+      "loss": 0.7372,
+      "step": 7855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.629084748383417e-07,
+      "loss": 0.5238,
+      "step": 7856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6178614821320466e-07,
+      "loss": 0.5485,
+      "step": 7857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.606651516290328e-07,
+      "loss": 0.5637,
+      "step": 7858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5954548524214326e-07,
+      "loss": 0.8108,
+      "step": 7859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.584271492086656e-07,
+      "loss": 0.6624,
+      "step": 7860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.573101436845462e-07,
+      "loss": 0.4808,
+      "step": 7861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.561944688255415e-07,
+      "loss": 0.6665,
+      "step": 7862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5508012478722495e-07,
+      "loss": 0.6544,
+      "step": 7863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5396711172498665e-07,
+      "loss": 0.8369,
+      "step": 7864
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.528554297940291e-07,
+      "loss": 0.9798,
+      "step": 7865
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5174507914936605e-07,
+      "loss": 0.5386,
+      "step": 7866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5063605994583036e-07,
+      "loss": 1.0417,
+      "step": 7867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.495283723380672e-07,
+      "loss": 0.6278,
+      "step": 7868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.484220164805375e-07,
+      "loss": 0.4986,
+      "step": 7869
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.473169925275134e-07,
+      "loss": 0.4728,
+      "step": 7870
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.462133006330838e-07,
+      "loss": 0.7239,
+      "step": 7871
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4511094095115115e-07,
+      "loss": 0.5891,
+      "step": 7872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4400991363543365e-07,
+      "loss": 0.5139,
+      "step": 7873
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.429102188394596e-07,
+      "loss": 0.4828,
+      "step": 7874
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4181185671657634e-07,
+      "loss": 0.5462,
+      "step": 7875
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4071482741994244e-07,
+      "loss": 0.6031,
+      "step": 7876
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3961913110253327e-07,
+      "loss": 0.6575,
+      "step": 7877
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.385247679171334e-07,
+      "loss": 0.5401,
+      "step": 7878
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3743173801634617e-07,
+      "loss": 0.797,
+      "step": 7879
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3634004155258645e-07,
+      "loss": 0.5935,
+      "step": 7880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3524967867808686e-07,
+      "loss": 0.5335,
+      "step": 7881
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.341606495448869e-07,
+      "loss": 0.5596,
+      "step": 7882
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.330729543048484e-07,
+      "loss": 0.7453,
+      "step": 7883
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3198659310964007e-07,
+      "loss": 0.547,
+      "step": 7884
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.309015661107518e-07,
+      "loss": 0.4949,
+      "step": 7885
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2981787345947936e-07,
+      "loss": 0.5775,
+      "step": 7886
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2873551530693837e-07,
+      "loss": 0.5693,
+      "step": 7887
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2765449180405707e-07,
+      "loss": 0.5141,
+      "step": 7888
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.265748031015771e-07,
+      "loss": 0.5561,
+      "step": 7889
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2549644935005243e-07,
+      "loss": 0.4664,
+      "step": 7890
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.244194306998528e-07,
+      "loss": 0.5247,
+      "step": 7891
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2334374730116147e-07,
+      "loss": 0.7976,
+      "step": 7892
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2226939930397617e-07,
+      "loss": 0.854,
+      "step": 7893
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2119638685810706e-07,
+      "loss": 0.511,
+      "step": 7894
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.201247101131778e-07,
+      "loss": 0.6934,
+      "step": 7895
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1905436921862995e-07,
+      "loss": 0.5275,
+      "step": 7896
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1798536432371086e-07,
+      "loss": 0.6682,
+      "step": 7897
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1691769557748896e-07,
+      "loss": 0.5795,
+      "step": 7898
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.15851363128843e-07,
+      "loss": 0.7494,
+      "step": 7899
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1478636712646845e-07,
+      "loss": 0.5816,
+      "step": 7900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.137227077188666e-07,
+      "loss": 0.8232,
+      "step": 7901
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.12660385054362e-07,
+      "loss": 0.5235,
+      "step": 7902
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.115993992810863e-07,
+      "loss": 0.5992,
+      "step": 7903
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.105397505469899e-07,
+      "loss": 0.6972,
+      "step": 7904
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.094814389998303e-07,
+      "loss": 0.5699,
+      "step": 7905
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0842446478718376e-07,
+      "loss": 0.532,
+      "step": 7906
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.073688280564381e-07,
+      "loss": 0.5494,
+      "step": 7907
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.063145289547965e-07,
+      "loss": 0.7074,
+      "step": 7908
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.052615676292704e-07,
+      "loss": 0.5341,
+      "step": 7909
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.042099442266911e-07,
+      "loss": 0.5943,
+      "step": 7910
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0315965889370033e-07,
+      "loss": 0.5724,
+      "step": 7911
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.02110711776752e-07,
+      "loss": 0.5472,
+      "step": 7912
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.010631030221168e-07,
+      "loss": 0.5995,
+      "step": 7913
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0001683277587466e-07,
+      "loss": 0.5881,
+      "step": 7914
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9897190118392436e-07,
+      "loss": 0.4541,
+      "step": 7915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.979283083919716e-07,
+      "loss": 0.6253,
+      "step": 7916
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9688605454553886e-07,
+      "loss": 0.5197,
+      "step": 7917
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.958451397899632e-07,
+      "loss": 0.7229,
+      "step": 7918
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9480556427039296e-07,
+      "loss": 0.4687,
+      "step": 7919
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9376732813178886e-07,
+      "loss": 0.5082,
+      "step": 7920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.927304315189262e-07,
+      "loss": 0.4845,
+      "step": 7921
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.916948745763938e-07,
+      "loss": 0.7221,
+      "step": 7922
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9066065744859493e-07,
+      "loss": 0.5745,
+      "step": 7923
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.89627780279741e-07,
+      "loss": 0.706,
+      "step": 7924
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.885962432138624e-07,
+      "loss": 0.5221,
+      "step": 7925
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.875660463947983e-07,
+      "loss": 0.5686,
+      "step": 7926
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.865371899662063e-07,
+      "loss": 0.4357,
+      "step": 7927
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8550967407155025e-07,
+      "loss": 0.6188,
+      "step": 7928
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8448349885410906e-07,
+      "loss": 0.5517,
+      "step": 7929
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.834586644569815e-07,
+      "loss": 0.6731,
+      "step": 7930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8243517102307004e-07,
+      "loss": 0.623,
+      "step": 7931
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8141301869509484e-07,
+      "loss": 0.5873,
+      "step": 7932
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.803922076155886e-07,
+      "loss": 0.6452,
+      "step": 7933
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7937273792689854e-07,
+      "loss": 0.5592,
+      "step": 7934
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7835460977117875e-07,
+      "loss": 0.5013,
+      "step": 7935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7733782329040345e-07,
+      "loss": 0.5466,
+      "step": 7936
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.763223786263559e-07,
+      "loss": 0.7352,
+      "step": 7937
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.753082759206339e-07,
+      "loss": 0.6621,
+      "step": 7938
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.742955153146466e-07,
+      "loss": 0.602,
+      "step": 7939
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.732840969496165e-07,
+      "loss": 0.5895,
+      "step": 7940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7227402096657985e-07,
+      "loss": 0.5694,
+      "step": 7941
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.71265287506386e-07,
+      "loss": 0.5505,
+      "step": 7942
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.702578967096937e-07,
+      "loss": 0.5884,
+      "step": 7943
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.692518487169794e-07,
+      "loss": 0.5265,
+      "step": 7944
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6824714366852863e-07,
+      "loss": 0.5384,
+      "step": 7945
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.672437817044416e-07,
+      "loss": 0.4772,
+      "step": 7946
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6624176296462755e-07,
+      "loss": 0.8242,
+      "step": 7947
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6524108758881573e-07,
+      "loss": 0.6285,
+      "step": 7948
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.642417557165434e-07,
+      "loss": 0.6636,
+      "step": 7949
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6324376748715694e-07,
+      "loss": 0.5037,
+      "step": 7950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6224712303982166e-07,
+      "loss": 0.7646,
+      "step": 7951
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6125182251351423e-07,
+      "loss": 0.6567,
+      "step": 7952
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.602578660470213e-07,
+      "loss": 0.7898,
+      "step": 7953
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5926525377894203e-07,
+      "loss": 0.9123,
+      "step": 7954
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5827398584769246e-07,
+      "loss": 0.5648,
+      "step": 7955
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.572840623914964e-07,
+      "loss": 0.6534,
+      "step": 7956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.562954835483945e-07,
+      "loss": 0.7036,
+      "step": 7957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.553082494562354e-07,
+      "loss": 0.6671,
+      "step": 7958
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.543223602526813e-07,
+      "loss": 0.5796,
+      "step": 7959
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.53337816075211e-07,
+      "loss": 0.5358,
+      "step": 7960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.523546170611114e-07,
+      "loss": 0.6418,
+      "step": 7961
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5137276334748285e-07,
+      "loss": 0.5825,
+      "step": 7962
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.50392255071238e-07,
+      "loss": 0.4726,
+      "step": 7963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.494130923691019e-07,
+      "loss": 0.6358,
+      "step": 7964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4843527537761547e-07,
+      "loss": 0.5728,
+      "step": 7965
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4745880423312397e-07,
+      "loss": 0.8163,
+      "step": 7966
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4648367907179406e-07,
+      "loss": 0.7043,
+      "step": 7967
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.455099000295992e-07,
+      "loss": 0.5224,
+      "step": 7968
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.445374672423252e-07,
+      "loss": 0.6263,
+      "step": 7969
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.435663808455725e-07,
+      "loss": 0.722,
+      "step": 7970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4259664097475276e-07,
+      "loss": 0.6368,
+      "step": 7971
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.416282477650901e-07,
+      "loss": 0.586,
+      "step": 7972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4066120135161973e-07,
+      "loss": 0.5934,
+      "step": 7973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.396955018691894e-07,
+      "loss": 0.5683,
+      "step": 7974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3873114945246033e-07,
+      "loss": 0.4927,
+      "step": 7975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3776814423590596e-07,
+      "loss": 0.5964,
+      "step": 7976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.368064863538101e-07,
+      "loss": 0.5757,
+      "step": 7977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.358461759402687e-07,
+      "loss": 0.5338,
+      "step": 7978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.348872131291914e-07,
+      "loss": 0.4767,
+      "step": 7979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3392959805430116e-07,
+      "loss": 0.6725,
+      "step": 7980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.32973330849129e-07,
+      "loss": 0.5613,
+      "step": 7981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3201841164701934e-07,
+      "loss": 0.7117,
+      "step": 7982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.310648405811312e-07,
+      "loss": 0.7292,
+      "step": 7983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.301126177844327e-07,
+      "loss": 0.481,
+      "step": 7984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.291617433897054e-07,
+      "loss": 0.6191,
+      "step": 7985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.282122175295421e-07,
+      "loss": 0.7079,
+      "step": 7986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2726404033634807e-07,
+      "loss": 0.8378,
+      "step": 7987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.263172119423419e-07,
+      "loss": 0.5776,
+      "step": 7988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2537173247955023e-07,
+      "loss": 0.7461,
+      "step": 7989
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2442760207981537e-07,
+      "loss": 0.5206,
+      "step": 7990
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2348482087478873e-07,
+      "loss": 0.7835,
+      "step": 7991
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.225433889959373e-07,
+      "loss": 0.5589,
+      "step": 7992
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.21603306574535e-07,
+      "loss": 0.4949,
+      "step": 7993
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2066457374167135e-07,
+      "loss": 0.5106,
+      "step": 7994
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.197271906282462e-07,
+      "loss": 0.6195,
+      "step": 7995
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.187911573649727e-07,
+      "loss": 0.5942,
+      "step": 7996
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.17856474082372e-07,
+      "loss": 0.529,
+      "step": 7997
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1692314091078093e-07,
+      "loss": 0.6064,
+      "step": 7998
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1599115798034654e-07,
+      "loss": 0.5883,
+      "step": 7999
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1506052542102817e-07,
+      "loss": 0.6704,
+      "step": 8000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1413124336259535e-07,
+      "loss": 0.7842,
+      "step": 8001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1320331193463115e-07,
+      "loss": 0.4881,
+      "step": 8002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.122767312665287e-07,
+      "loss": 0.6401,
+      "step": 8003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.113515014874935e-07,
+      "loss": 0.6124,
+      "step": 8004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.104276227265446e-07,
+      "loss": 0.67,
+      "step": 8005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.095050951125089e-07,
+      "loss": 0.5718,
+      "step": 8006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0858391877402914e-07,
+      "loss": 0.8053,
+      "step": 8007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0766409383955366e-07,
+      "loss": 0.4924,
+      "step": 8008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.067456204373476e-07,
+      "loss": 0.9126,
+      "step": 8009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0582849869548735e-07,
+      "loss": 0.5552,
+      "step": 8010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.049127287418585e-07,
+      "loss": 0.538,
+      "step": 8011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0399831070415774e-07,
+      "loss": 0.8794,
+      "step": 8012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.030852447098964e-07,
+      "loss": 0.7983,
+      "step": 8013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.02173530886396e-07,
+      "loss": 0.7204,
+      "step": 8014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.012631693607881e-07,
+      "loss": 0.6873,
+      "step": 8015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0035416026001573e-07,
+      "loss": 0.5402,
+      "step": 8016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9944650371083404e-07,
+      "loss": 0.6405,
+      "step": 8017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9854019983981184e-07,
+      "loss": 0.6459,
+      "step": 8018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.976352487733269e-07,
+      "loss": 0.9604,
+      "step": 8019
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9673165063756725e-07,
+      "loss": 0.6608,
+      "step": 8020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.95829405558532e-07,
+      "loss": 0.557,
+      "step": 8021
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.949285136620383e-07,
+      "loss": 0.5445,
+      "step": 8022
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.940289750737069e-07,
+      "loss": 0.7083,
+      "step": 8023
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.931307899189717e-07,
+      "loss": 0.4953,
+      "step": 8024
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.922339583230793e-07,
+      "loss": 0.5327,
+      "step": 8025
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.913384804110875e-07,
+      "loss": 0.538,
+      "step": 8026
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9044435630786516e-07,
+      "loss": 0.8079,
+      "step": 8027
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.895515861380893e-07,
+      "loss": 0.7442,
+      "step": 8028
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.886601700262537e-07,
+      "loss": 0.546,
+      "step": 8029
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.877701080966611e-07,
+      "loss": 0.7519,
+      "step": 8030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8688140047342104e-07,
+      "loss": 0.5729,
+      "step": 8031
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.859940472804612e-07,
+      "loss": 0.584,
+      "step": 8032
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8510804864151475e-07,
+      "loss": 0.5258,
+      "step": 8033
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.842234046801318e-07,
+      "loss": 0.6678,
+      "step": 8034
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8334011551966577e-07,
+      "loss": 0.7006,
+      "step": 8035
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.824581812832883e-07,
+      "loss": 0.6,
+      "step": 8036
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8157760209397867e-07,
+      "loss": 0.9694,
+      "step": 8037
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.806983780745287e-07,
+      "loss": 0.7503,
+      "step": 8038
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.798205093475381e-07,
+      "loss": 0.7413,
+      "step": 8039
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.789439960354212e-07,
+      "loss": 0.601,
+      "step": 8040
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.780688382604035e-07,
+      "loss": 0.6739,
+      "step": 8041
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.771950361445175e-07,
+      "loss": 0.5563,
+      "step": 8042
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7632258980961e-07,
+      "loss": 0.4939,
+      "step": 8043
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.754514993773383e-07,
+      "loss": 0.9273,
+      "step": 8044
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.7458176496917066e-07,
+      "loss": 0.5193,
+      "step": 8045
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.737133867063846e-07,
+      "loss": 0.5197,
+      "step": 8046
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7284636471006984e-07,
+      "loss": 0.5652,
+      "step": 8047
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7198069910112755e-07,
+      "loss": 0.8747,
+      "step": 8048
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7111639000027003e-07,
+      "loss": 0.5507,
+      "step": 8049
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7025343752801636e-07,
+      "loss": 0.5575,
+      "step": 8050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.693918418047026e-07,
+      "loss": 0.8357,
+      "step": 8051
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.685316029504703e-07,
+      "loss": 0.6146,
+      "step": 8052
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.67672721085277e-07,
+      "loss": 0.512,
+      "step": 8053
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.668151963288845e-07,
+      "loss": 0.8438,
+      "step": 8054
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6595902880087177e-07,
+      "loss": 0.573,
+      "step": 8055
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.651042186206243e-07,
+      "loss": 0.603,
+      "step": 8056
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6425076590734124e-07,
+      "loss": 0.7487,
+      "step": 8057
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6339867078002844e-07,
+      "loss": 0.5226,
+      "step": 8058
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.625479333575065e-07,
+      "loss": 0.5644,
+      "step": 8059
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6169855375840716e-07,
+      "loss": 0.6696,
+      "step": 8060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6085053210116674e-07,
+      "loss": 0.8335,
+      "step": 8061
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6000386850403844e-07,
+      "loss": 0.5615,
+      "step": 8062
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.591585630850835e-07,
+      "loss": 0.5762,
+      "step": 8063
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.583146159621752e-07,
+      "loss": 0.5479,
+      "step": 8064
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.574720272529951e-07,
+      "loss": 0.7968,
+      "step": 8065
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5663079707503703e-07,
+      "loss": 0.4779,
+      "step": 8066
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.557909255456048e-07,
+      "loss": 0.8125,
+      "step": 8067
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.549524127818137e-07,
+      "loss": 0.5659,
+      "step": 8068
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.54115258900588e-07,
+      "loss": 0.4872,
+      "step": 8069
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.532794640186631e-07,
+      "loss": 0.763,
+      "step": 8070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.524450282525859e-07,
+      "loss": 0.5475,
+      "step": 8071
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.516119517187132e-07,
+      "loss": 0.8093,
+      "step": 8072
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.507802345332111e-07,
+      "loss": 0.7724,
+      "step": 8073
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4994987681205674e-07,
+      "loss": 0.5244,
+      "step": 8074
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.491208786710386e-07,
+      "loss": 0.8491,
+      "step": 8075
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.482932402257554e-07,
+      "loss": 0.5956,
+      "step": 8076
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.474669615916159e-07,
+      "loss": 0.5869,
+      "step": 8077
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4664204288383784e-07,
+      "loss": 0.7479,
+      "step": 8078
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.458184842174538e-07,
+      "loss": 0.5787,
+      "step": 8079
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.449962857072996e-07,
+      "loss": 0.5178,
+      "step": 8080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.44175447468028e-07,
+      "loss": 0.8986,
+      "step": 8081
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.433559696140986e-07,
+      "loss": 0.6187,
+      "step": 8082
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.425378522597832e-07,
+      "loss": 0.7244,
+      "step": 8083
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4172109551916065e-07,
+      "loss": 0.7973,
+      "step": 8084
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.409056995061243e-07,
+      "loss": 0.4942,
+      "step": 8085
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4009166433437534e-07,
+      "loss": 0.6395,
+      "step": 8086
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.392789901174253e-07,
+      "loss": 0.6828,
+      "step": 8087
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3846767696859565e-07,
+      "loss": 0.6786,
+      "step": 8088
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3765772500102036e-07,
+      "loss": 0.6939,
+      "step": 8089
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.368491343276391e-07,
+      "loss": 0.5374,
+      "step": 8090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.360419050612084e-07,
+      "loss": 0.6585,
+      "step": 8091
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.35236037314287e-07,
+      "loss": 0.4943,
+      "step": 8092
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3443153119925067e-07,
+      "loss": 0.6206,
+      "step": 8093
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.336283868282807e-07,
+      "loss": 0.5799,
+      "step": 8094
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.32826604313372e-07,
+      "loss": 0.5794,
+      "step": 8095
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3202618376632624e-07,
+      "loss": 0.8106,
+      "step": 8096
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3122712529875745e-07,
+      "loss": 0.8177,
+      "step": 8097
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3042942902208988e-07,
+      "loss": 0.6217,
+      "step": 8098
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.296330950475556e-07,
+      "loss": 0.7739,
+      "step": 8099
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2883812348619916e-07,
+      "loss": 0.4993,
+      "step": 8100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.28044514448873e-07,
+      "loss": 0.5645,
+      "step": 8101
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2725226804624302e-07,
+      "loss": 0.4862,
+      "step": 8102
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2646138438877973e-07,
+      "loss": 0.7627,
+      "step": 8103
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.256718635867672e-07,
+      "loss": 0.5341,
+      "step": 8104
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.248837057503006e-07,
+      "loss": 0.5607,
+      "step": 8105
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.240969109892832e-07,
+      "loss": 0.6631,
+      "step": 8106
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.233114794134261e-07,
+      "loss": 0.6034,
+      "step": 8107
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2252741113225506e-07,
+      "loss": 0.612,
+      "step": 8108
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2174470625510147e-07,
+      "loss": 0.7841,
+      "step": 8109
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2096336489111025e-07,
+      "loss": 0.576,
+      "step": 8110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2018338714923205e-07,
+      "loss": 0.6011,
+      "step": 8111
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1940477313822984e-07,
+      "loss": 0.4918,
+      "step": 8112
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1862752296667678e-07,
+      "loss": 0.6409,
+      "step": 8113
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.178516367429584e-07,
+      "loss": 0.6535,
+      "step": 8114
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.170771145752615e-07,
+      "loss": 0.7865,
+      "step": 8115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1630395657159186e-07,
+      "loss": 0.6013,
+      "step": 8116
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1553216283975887e-07,
+      "loss": 0.5746,
+      "step": 8117
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1476173348738748e-07,
+      "loss": 0.7201,
+      "step": 8118
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1399266862190516e-07,
+      "loss": 0.6522,
+      "step": 8119
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1322496835055384e-07,
+      "loss": 0.8872,
+      "step": 8120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.124586327803857e-07,
+      "loss": 0.654,
+      "step": 8121
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.116936620182608e-07,
+      "loss": 0.5529,
+      "step": 8122
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1093005617084828e-07,
+      "loss": 0.522,
+      "step": 8123
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1016781534462848e-07,
+      "loss": 0.5507,
+      "step": 8124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0940693964589087e-07,
+      "loss": 0.8465,
+      "step": 8125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.086474291807361e-07,
+      "loss": 0.8627,
+      "step": 8126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0788928405506946e-07,
+      "loss": 0.7359,
+      "step": 8127
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0713250437461197e-07,
+      "loss": 0.5697,
+      "step": 8128
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0637709024489028e-07,
+      "loss": 0.5534,
+      "step": 8129
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.056230417712435e-07,
+      "loss": 0.6632,
+      "step": 8130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0487035905881525e-07,
+      "loss": 0.6178,
+      "step": 8131
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0411904221256606e-07,
+      "loss": 0.5435,
+      "step": 8132
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.03369091337261e-07,
+      "loss": 0.5257,
+      "step": 8133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0262050653747423e-07,
+      "loss": 0.6501,
+      "step": 8134
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.018732879175933e-07,
+      "loss": 0.8666,
+      "step": 8135
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0112743558181048e-07,
+      "loss": 0.6474,
+      "step": 8136
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0038294963413251e-07,
+      "loss": 0.6404,
+      "step": 8137
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9963983017837084e-07,
+      "loss": 0.5591,
+      "step": 8138
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9889807731814925e-07,
+      "loss": 0.5101,
+      "step": 8139
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9815769115690053e-07,
+      "loss": 0.61,
+      "step": 8140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9741867179786655e-07,
+      "loss": 0.5504,
+      "step": 8141
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9668101934409933e-07,
+      "loss": 0.6909,
+      "step": 8142
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.959447338984577e-07,
+      "loss": 0.7001,
+      "step": 8143
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9520981556361395e-07,
+      "loss": 0.5764,
+      "step": 8144
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9447626444204726e-07,
+      "loss": 0.5731,
+      "step": 8145
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9374408063604577e-07,
+      "loss": 0.5389,
+      "step": 8146
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9301326424770673e-07,
+      "loss": 0.7972,
+      "step": 8147
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.922838153789397e-07,
+      "loss": 0.5601,
+      "step": 8148
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9155573413146111e-07,
+      "loss": 0.5614,
+      "step": 8149
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.908290206067953e-07,
+      "loss": 0.9197,
+      "step": 8150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.90103674906279e-07,
+      "loss": 0.8193,
+      "step": 8151
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8937969713105797e-07,
+      "loss": 0.4738,
+      "step": 8152
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8865708738208477e-07,
+      "loss": 0.6613,
+      "step": 8153
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8793584576012214e-07,
+      "loss": 0.6684,
+      "step": 8154
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8721597236574407e-07,
+      "loss": 0.546,
+      "step": 8155
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.864974672993314e-07,
+      "loss": 0.5897,
+      "step": 8156
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8578033066107392e-07,
+      "loss": 0.6328,
+      "step": 8157
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8506456255097171e-07,
+      "loss": 0.5209,
+      "step": 8158
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8435016306883603e-07,
+      "loss": 0.4646,
+      "step": 8159
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8363713231428381e-07,
+      "loss": 0.4916,
+      "step": 8160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8292547038674112e-07,
+      "loss": 0.6451,
+      "step": 8161
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8221517738544637e-07,
+      "loss": 0.7256,
+      "step": 8162
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8150625340944472e-07,
+      "loss": 0.5816,
+      "step": 8163
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8079869855759268e-07,
+      "loss": 0.6076,
+      "step": 8164
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8009251292855023e-07,
+      "loss": 0.5151,
+      "step": 8165
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7938769662079301e-07,
+      "loss": 0.5779,
+      "step": 8166
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7868424973260135e-07,
+      "loss": 0.5801,
+      "step": 8167
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7798217236206895e-07,
+      "loss": 0.5635,
+      "step": 8168
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7728146460709305e-07,
+      "loss": 0.7185,
+      "step": 8169
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7658212656538443e-07,
+      "loss": 0.5626,
+      "step": 8170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.758841583344606e-07,
+      "loss": 0.5049,
+      "step": 8171
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7518756001164817e-07,
+      "loss": 0.5242,
+      "step": 8172
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7449233169408498e-07,
+      "loss": 0.6066,
+      "step": 8173
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7379847347871348e-07,
+      "loss": 0.7186,
+      "step": 8174
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7310598546228963e-07,
+      "loss": 0.5296,
+      "step": 8175
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7241486774137506e-07,
+      "loss": 0.5513,
+      "step": 8176
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7172512041234156e-07,
+      "loss": 0.583,
+      "step": 8177
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7103674357137112e-07,
+      "loss": 0.7719,
+      "step": 8178
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7034973731445247e-07,
+      "loss": 0.8518,
+      "step": 8179
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6966410173738456e-07,
+      "loss": 0.529,
+      "step": 8180
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6897983693577313e-07,
+      "loss": 0.7023,
+      "step": 8181
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.682969430050363e-07,
+      "loss": 0.6789,
+      "step": 8182
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.676154200404001e-07,
+      "loss": 0.8582,
+      "step": 8183
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.669352681368941e-07,
+      "loss": 0.5241,
+      "step": 8184
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6625648738936463e-07,
+      "loss": 0.8356,
+      "step": 8185
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6557907789246262e-07,
+      "loss": 0.5609,
+      "step": 8186
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6490303974064814e-07,
+      "loss": 0.6549,
+      "step": 8187
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6422837302818905e-07,
+      "loss": 0.4485,
+      "step": 8188
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6355507784916458e-07,
+      "loss": 0.7289,
+      "step": 8189
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6288315429746182e-07,
+      "loss": 0.5579,
+      "step": 8190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.622126024667736e-07,
+      "loss": 0.5623,
+      "step": 8191
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6154342245060628e-07,
+      "loss": 0.6417,
+      "step": 8192
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6087561434227185e-07,
+      "loss": 0.523,
+      "step": 8193
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6020917823489247e-07,
+      "loss": 0.6307,
+      "step": 8194
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5954411422139716e-07,
+      "loss": 0.7343,
+      "step": 8195
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5888042239452394e-07,
+      "loss": 0.766,
+      "step": 8196
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5821810284682216e-07,
+      "loss": 0.8338,
+      "step": 8197
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5755715567064788e-07,
+      "loss": 0.7786,
+      "step": 8198
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.56897580958163e-07,
+      "loss": 0.5161,
+      "step": 8199
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5623937880134388e-07,
+      "loss": 0.5293,
+      "step": 8200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.555825492919716e-07,
+      "loss": 0.5681,
+      "step": 8201
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5492709252163728e-07,
+      "loss": 0.5382,
+      "step": 8202
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.54273008581739e-07,
+      "loss": 0.6877,
+      "step": 8203
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5362029756348373e-07,
+      "loss": 0.5034,
+      "step": 8204
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5296895955788982e-07,
+      "loss": 0.4865,
+      "step": 8205
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5231899465578125e-07,
+      "loss": 0.5811,
+      "step": 8206
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5167040294779001e-07,
+      "loss": 0.7507,
+      "step": 8207
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5102318452435926e-07,
+      "loss": 0.5636,
+      "step": 8208
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.503773394757402e-07,
+      "loss": 0.5849,
+      "step": 8209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.497328678919896e-07,
+      "loss": 0.5191,
+      "step": 8210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4908976986297562e-07,
+      "loss": 0.7324,
+      "step": 8211
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4844804547837545e-07,
+      "loss": 0.5426,
+      "step": 8212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4780769482767078e-07,
+      "loss": 0.6346,
+      "step": 8213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4716871800015574e-07,
+      "loss": 0.4785,
+      "step": 8214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4653111508493133e-07,
+      "loss": 0.7385,
+      "step": 8215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.458948861709075e-07,
+      "loss": 0.5403,
+      "step": 8216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4526003134680112e-07,
+      "loss": 0.7073,
+      "step": 8217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4462655070113906e-07,
+      "loss": 0.5738,
+      "step": 8218
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4399444432225518e-07,
+      "loss": 0.8332,
+      "step": 8219
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4336371229829448e-07,
+      "loss": 0.6067,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4273435471720665e-07,
+      "loss": 0.9188,
+      "step": 8221
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4210637166675256e-07,
+      "loss": 0.6574,
+      "step": 8222
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4147976323449774e-07,
+      "loss": 0.7793,
+      "step": 8223
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.408545295078223e-07,
+      "loss": 0.6169,
+      "step": 8224
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4023067057390982e-07,
+      "loss": 0.5453,
+      "step": 8225
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3960818651975183e-07,
+      "loss": 0.8526,
+      "step": 8226
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3898707743215111e-07,
+      "loss": 0.6103,
+      "step": 8227
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3836734339771618e-07,
+      "loss": 0.5033,
+      "step": 8228
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3774898450286678e-07,
+      "loss": 0.5726,
+      "step": 8229
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3713200083382616e-07,
+      "loss": 0.6351,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3651639247662996e-07,
+      "loss": 0.6367,
+      "step": 8231
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.359021595171206e-07,
+      "loss": 0.6451,
+      "step": 8232
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.352893020409507e-07,
+      "loss": 0.8321,
+      "step": 8233
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.346778201335752e-07,
+      "loss": 0.6892,
+      "step": 8234
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3406771388026485e-07,
+      "loss": 0.6085,
+      "step": 8235
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3345898336609264e-07,
+      "loss": 0.5109,
+      "step": 8236
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3285162867594402e-07,
+      "loss": 0.843,
+      "step": 8237
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3224564989450907e-07,
+      "loss": 0.5791,
+      "step": 8238
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3164104710628788e-07,
+      "loss": 0.5049,
+      "step": 8239
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3103782039558865e-07,
+      "loss": 0.5214,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.304359698465274e-07,
+      "loss": 0.4869,
+      "step": 8241
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.29835495543027e-07,
+      "loss": 0.5749,
+      "step": 8242
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2923639756882044e-07,
+      "loss": 0.4696,
+      "step": 8243
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2863867600744874e-07,
+      "loss": 0.5175,
+      "step": 8244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.280423309422596e-07,
+      "loss": 0.5834,
+      "step": 8245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2744736245640987e-07,
+      "loss": 0.5923,
+      "step": 8246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.268537706328632e-07,
+      "loss": 0.6019,
+      "step": 8247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2626155555439225e-07,
+      "loss": 0.4378,
+      "step": 8248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.256707173035776e-07,
+      "loss": 0.9479,
+      "step": 8249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2508125596280785e-07,
+      "loss": 0.507,
+      "step": 8250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2449317161427942e-07,
+      "loss": 0.6113,
+      "step": 8251
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2390646433999676e-07,
+      "loss": 0.5357,
+      "step": 8252
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2332113422177216e-07,
+      "loss": 0.6103,
+      "step": 8253
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2273718134122592e-07,
+      "loss": 0.5683,
+      "step": 8254
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2215460577978734e-07,
+      "loss": 0.5513,
+      "step": 8255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2157340761869253e-07,
+      "loss": 0.6384,
+      "step": 8256
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2099358693898445e-07,
+      "loss": 0.5176,
+      "step": 8257
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.204151438215162e-07,
+      "loss": 0.8161,
+      "step": 8258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1983807834694772e-07,
+      "loss": 0.664,
+      "step": 8259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1926239059574796e-07,
+      "loss": 0.5252,
+      "step": 8260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1868808064819049e-07,
+      "loss": 0.5719,
+      "step": 8261
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1811514858436124e-07,
+      "loss": 0.4925,
+      "step": 8262
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1754359448415186e-07,
+      "loss": 0.6376,
+      "step": 8263
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1697341842725973e-07,
+      "loss": 0.5174,
+      "step": 8264
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1640462049319346e-07,
+      "loss": 0.7469,
+      "step": 8265
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1583720076126847e-07,
+      "loss": 0.5815,
+      "step": 8266
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1527115931060706e-07,
+      "loss": 0.6512,
+      "step": 8267
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1470649622014052e-07,
+      "loss": 0.5373,
+      "step": 8268
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1414321156860697e-07,
+      "loss": 0.5694,
+      "step": 8269
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1358130543455248e-07,
+      "loss": 0.8636,
+      "step": 8270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1302077789633215e-07,
+      "loss": 0.5225,
+      "step": 8271
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1246162903210567e-07,
+      "loss": 0.5826,
+      "step": 8272
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1190385891984513e-07,
+      "loss": 0.6655,
+      "step": 8273
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1134746763732717e-07,
+      "loss": 0.5524,
+      "step": 8274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1079245526213645e-07,
+      "loss": 0.5336,
+      "step": 8275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1023882187166657e-07,
+      "loss": 0.5269,
+      "step": 8276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.096865675431169e-07,
+      "loss": 0.6082,
+      "step": 8277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0913569235349586e-07,
+      "loss": 0.5787,
+      "step": 8278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0858619637961976e-07,
+      "loss": 0.8079,
+      "step": 8279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0803807969811287e-07,
+      "loss": 0.6617,
+      "step": 8280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0749134238540516e-07,
+      "loss": 0.5083,
+      "step": 8281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0694598451773674e-07,
+      "loss": 0.6089,
+      "step": 8282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0640200617115237e-07,
+      "loss": 0.6852,
+      "step": 8283
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0585940742150802e-07,
+      "loss": 0.538,
+      "step": 8284
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0531818834446539e-07,
+      "loss": 0.7276,
+      "step": 8285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.04778349015493e-07,
+      "loss": 0.488,
+      "step": 8286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.042398895098684e-07,
+      "loss": 0.5377,
+      "step": 8287
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.03702809902676e-07,
+      "loss": 0.7781,
+      "step": 8288
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.031671102688081e-07,
+      "loss": 0.9977,
+      "step": 8289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0263279068296495e-07,
+      "loss": 0.6514,
+      "step": 8290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0209985121965249e-07,
+      "loss": 0.6575,
+      "step": 8291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0156829195318685e-07,
+      "loss": 0.6302,
+      "step": 8292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0103811295769095e-07,
+      "loss": 0.5275,
+      "step": 8293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0050931430709343e-07,
+      "loss": 0.5341,
+      "step": 8294
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.998189607513198e-08,
+      "loss": 0.5874,
+      "step": 8295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.94558583353522e-08,
+      "loss": 0.5553,
+      "step": 8296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.893120116110656e-08,
+      "loss": 0.5876,
+      "step": 8297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.840792462555426e-08,
+      "loss": 0.6281,
+      "step": 8298
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.788602880166253e-08,
+      "loss": 0.6504,
+      "step": 8299
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.73655137622087e-08,
+      "loss": 0.7463,
+      "step": 8300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.68463795797725e-08,
+      "loss": 0.8246,
+      "step": 8301
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.632862632674489e-08,
+      "loss": 0.5189,
+      "step": 8302
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.581225407532368e-08,
+      "loss": 0.5555,
+      "step": 8303
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.52972628975124e-08,
+      "loss": 0.5527,
+      "step": 8304
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.478365286512359e-08,
+      "loss": 0.8183,
+      "step": 8305
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.427142404977773e-08,
+      "loss": 0.476,
+      "step": 8306
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.376057652289993e-08,
+      "loss": 0.7519,
+      "step": 8307
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.325111035572543e-08,
+      "loss": 0.6035,
+      "step": 8308
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.274302561929626e-08,
+      "loss": 0.5481,
+      "step": 8309
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.223632238446134e-08,
+      "loss": 0.4497,
+      "step": 8310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.173100072187635e-08,
+      "loss": 0.6037,
+      "step": 8311
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.122706070200604e-08,
+      "loss": 0.6415,
+      "step": 8312
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.072450239512088e-08,
+      "loss": 0.5397,
+      "step": 8313
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.022332587130034e-08,
+      "loss": 0.6054,
+      "step": 8314
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.972353120042853e-08,
+      "loss": 0.7698,
+      "step": 8315
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.922511845219972e-08,
+      "loss": 0.6609,
+      "step": 8316
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.872808769611274e-08,
+      "loss": 0.6241,
+      "step": 8317
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.823243900147771e-08,
+      "loss": 0.5045,
+      "step": 8318
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.773817243740823e-08,
+      "loss": 0.6985,
+      "step": 8319
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.724528807282695e-08,
+      "loss": 0.5576,
+      "step": 8320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.675378597646223e-08,
+      "loss": 0.8485,
+      "step": 8321
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.626366621685145e-08,
+      "loss": 0.5655,
+      "step": 8322
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.577492886233885e-08,
+      "loss": 0.6467,
+      "step": 8323
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.528757398107656e-08,
+      "loss": 0.8294,
+      "step": 8324
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.480160164102024e-08,
+      "loss": 0.666,
+      "step": 8325
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.43170119099379e-08,
+      "loss": 0.6048,
+      "step": 8326
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.383380485540216e-08,
+      "loss": 0.6543,
+      "step": 8327
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.335198054479244e-08,
+      "loss": 0.5883,
+      "step": 8328
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.287153904529499e-08,
+      "loss": 0.6374,
+      "step": 8329
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.239248042390624e-08,
+      "loss": 0.526,
+      "step": 8330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.191480474742607e-08,
+      "loss": 0.5759,
+      "step": 8331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.143851208246456e-08,
+      "loss": 0.9146,
+      "step": 8332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.096360249543633e-08,
+      "loss": 0.7889,
+      "step": 8333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.049007605256509e-08,
+      "loss": 0.843,
+      "step": 8334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.001793281988135e-08,
+      "loss": 0.754,
+      "step": 8335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.954717286322133e-08,
+      "loss": 0.5521,
+      "step": 8336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.907779624823031e-08,
+      "loss": 0.82,
+      "step": 8337
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.860980304035925e-08,
+      "loss": 0.5008,
+      "step": 8338
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.814319330486709e-08,
+      "loss": 0.8543,
+      "step": 8339
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.767796710681951e-08,
+      "loss": 0.6708,
+      "step": 8340
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.721412451108911e-08,
+      "loss": 0.7609,
+      "step": 8341
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.675166558235526e-08,
+      "loss": 0.526,
+      "step": 8342
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.629059038510522e-08,
+      "loss": 0.5224,
+      "step": 8343
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.583089898363204e-08,
+      "loss": 0.6993,
+      "step": 8344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.53725914420378e-08,
+      "loss": 0.5692,
+      "step": 8345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.491566782423022e-08,
+      "loss": 0.694,
+      "step": 8346
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.44601281939239e-08,
+      "loss": 0.6108,
+      "step": 8347
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.400597261464248e-08,
+      "loss": 0.5844,
+      "step": 8348
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.355320114971199e-08,
+      "loss": 0.5921,
+      "step": 8349
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.310181386226967e-08,
+      "loss": 0.567,
+      "step": 8350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.265181081525962e-08,
+      "loss": 0.938,
+      "step": 8351
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.220319207143056e-08,
+      "loss": 0.4889,
+      "step": 8352
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.17559576933391e-08,
+      "loss": 0.4855,
+      "step": 8353
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.13101077433509e-08,
+      "loss": 0.8829,
+      "step": 8354
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.086564228363624e-08,
+      "loss": 0.6538,
+      "step": 8355
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.04225613761711e-08,
+      "loss": 0.5815,
+      "step": 8356
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.99808650827416e-08,
+      "loss": 0.5787,
+      "step": 8357
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.95405534649396e-08,
+      "loss": 0.5465,
+      "step": 8358
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.910162658416375e-08,
+      "loss": 0.6055,
+      "step": 8359
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.866408450161843e-08,
+      "loss": 0.6711,
+      "step": 8360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.822792727831706e-08,
+      "loss": 0.5566,
+      "step": 8361
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.779315497507765e-08,
+      "loss": 0.5507,
+      "step": 8362
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.735976765252838e-08,
+      "loss": 0.592,
+      "step": 8363
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.69277653710998e-08,
+      "loss": 0.6944,
+      "step": 8364
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.64971481910337e-08,
+      "loss": 0.5634,
+      "step": 8365
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.60679161723754e-08,
+      "loss": 0.7628,
+      "step": 8366
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.564006937497925e-08,
+      "loss": 0.5191,
+      "step": 8367
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.521360785850639e-08,
+      "loss": 0.5525,
+      "step": 8368
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.478853168242261e-08,
+      "loss": 0.7078,
+      "step": 8369
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.436484090600159e-08,
+      "loss": 0.6045,
+      "step": 8370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.394253558832608e-08,
+      "loss": 0.5484,
+      "step": 8371
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.352161578828341e-08,
+      "loss": 0.7487,
+      "step": 8372
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.310208156456666e-08,
+      "loss": 0.8062,
+      "step": 8373
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.268393297567899e-08,
+      "loss": 0.5807,
+      "step": 8374
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.22671700799271e-08,
+      "loss": 0.5609,
+      "step": 8375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.185179293542564e-08,
+      "loss": 0.6098,
+      "step": 8376
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.143780160009716e-08,
+      "loss": 0.9446,
+      "step": 8377
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.102519613166991e-08,
+      "loss": 0.9796,
+      "step": 8378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.061397658767898e-08,
+      "loss": 0.5946,
+      "step": 8379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.020414302546518e-08,
+      "loss": 0.8556,
+      "step": 8380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979569550217834e-08,
+      "loss": 0.5885,
+      "step": 8381
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.938863407477402e-08,
+      "loss": 0.5846,
+      "step": 8382
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.898295880001237e-08,
+      "loss": 0.5465,
+      "step": 8383
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.8578669734463686e-08,
+      "loss": 0.5968,
+      "step": 8384
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.8175766934502886e-08,
+      "loss": 0.5717,
+      "step": 8385
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7774250456312794e-08,
+      "loss": 0.4684,
+      "step": 8386
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.737412035588086e-08,
+      "loss": 0.8148,
+      "step": 8387
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.6975376689003545e-08,
+      "loss": 0.839,
+      "step": 8388
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.657801951128306e-08,
+      "loss": 0.6012,
+      "step": 8389
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.6182048878128395e-08,
+      "loss": 0.482,
+      "step": 8390
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.578746484475428e-08,
+      "loss": 0.7964,
+      "step": 8391
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.539426746618337e-08,
+      "loss": 0.4949,
+      "step": 8392
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.500245679724403e-08,
+      "loss": 0.7434,
+      "step": 8393
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.461203289257255e-08,
+      "loss": 0.5773,
+      "step": 8394
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.422299580660983e-08,
+      "loss": 0.6327,
+      "step": 8395
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.383534559360582e-08,
+      "loss": 0.8939,
+      "step": 8396
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.344908230761503e-08,
+      "loss": 0.5945,
+      "step": 8397
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3064206002498844e-08,
+      "loss": 0.5477,
+      "step": 8398
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.268071673192654e-08,
+      "loss": 0.8463,
+      "step": 8399
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.229861454937313e-08,
+      "loss": 0.5688,
+      "step": 8400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.191789950812043e-08,
+      "loss": 0.5574,
+      "step": 8401
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.1538571661254867e-08,
+      "loss": 0.5703,
+      "step": 8402
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.1160631061673014e-08,
+      "loss": 0.5383,
+      "step": 8403
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0784077762076054e-08,
+      "loss": 0.5029,
+      "step": 8404
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.040891181497198e-08,
+      "loss": 0.6938,
+      "step": 8405
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0035133272673396e-08,
+      "loss": 0.777,
+      "step": 8406
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.966274218730305e-08,
+      "loss": 0.5396,
+      "step": 8407
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.929173861078829e-08,
+      "loss": 0.7396,
+      "step": 8408
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8922122594863283e-08,
+      "loss": 0.6267,
+      "step": 8409
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.855389419106682e-08,
+      "loss": 0.5414,
+      "step": 8410
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.818705345074781e-08,
+      "loss": 0.802,
+      "step": 8411
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.782160042505979e-08,
+      "loss": 0.778,
+      "step": 8412
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.745753516496088e-08,
+      "loss": 0.6482,
+      "step": 8413
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.7094857721219355e-08,
+      "loss": 0.8406,
+      "step": 8414
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.67335681444081e-08,
+      "loss": 0.5885,
+      "step": 8415
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.637366648490571e-08,
+      "loss": 0.5096,
+      "step": 8416
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6015152792898695e-08,
+      "loss": 0.5387,
+      "step": 8417
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.56580271183793e-08,
+      "loss": 0.6248,
+      "step": 8418
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.530228951114546e-08,
+      "loss": 0.5384,
+      "step": 8419
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.494794002080416e-08,
+      "loss": 0.5732,
+      "step": 8420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.459497869676588e-08,
+      "loss": 0.7163,
+      "step": 8421
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.4243405588249024e-08,
+      "loss": 0.639,
+      "step": 8422
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.389322074427771e-08,
+      "loss": 0.5725,
+      "step": 8423
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.354442421368399e-08,
+      "loss": 0.4566,
+      "step": 8424
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.319701604510451e-08,
+      "loss": 0.6245,
+      "step": 8425
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.285099628698164e-08,
+      "loss": 0.5867,
+      "step": 8426
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2506364987568995e-08,
+      "loss": 0.4848,
+      "step": 8427
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.216312219492036e-08,
+      "loss": 0.6314,
+      "step": 8428
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.182126795689967e-08,
+      "loss": 0.5939,
+      "step": 8429
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.148080232117546e-08,
+      "loss": 0.6079,
+      "step": 8430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.114172533522531e-08,
+      "loss": 0.5922,
+      "step": 8431
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.080403704632918e-08,
+      "loss": 0.5709,
+      "step": 8432
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0467737501576066e-08,
+      "loss": 0.6137,
+      "step": 8433
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0132826747861784e-08,
+      "loss": 0.485,
+      "step": 8434
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.979930483188676e-08,
+      "loss": 0.7138,
+      "step": 8435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9467171800157135e-08,
+      "loss": 0.6833,
+      "step": 8436
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9136427698988066e-08,
+      "loss": 0.5995,
+      "step": 8437
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8807072574499337e-08,
+      "loss": 0.517,
+      "step": 8438
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.847910647261754e-08,
+      "loss": 0.5089,
+      "step": 8439
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.815252943907499e-08,
+      "loss": 0.5282,
+      "step": 8440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.782734151941192e-08,
+      "loss": 0.5324,
+      "step": 8441
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7503542758970944e-08,
+      "loss": 0.5193,
+      "step": 8442
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.718113320290706e-08,
+      "loss": 0.6761,
+      "step": 8443
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.686011289617431e-08,
+      "loss": 0.6212,
+      "step": 8444
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.65404818835402e-08,
+      "loss": 0.8941,
+      "step": 8445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.6222240209574654e-08,
+      "loss": 0.6874,
+      "step": 8446
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.590538791865328e-08,
+      "loss": 0.8077,
+      "step": 8447
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5589925054958504e-08,
+      "loss": 0.5044,
+      "step": 8448
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.527585166248182e-08,
+      "loss": 0.5194,
+      "step": 8449
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.496316778501596e-08,
+      "loss": 0.6006,
+      "step": 8450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.465187346616494e-08,
+      "loss": 0.9707,
+      "step": 8451
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4341968749335155e-08,
+      "loss": 0.5117,
+      "step": 8452
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4033453677742024e-08,
+      "loss": 0.872,
+      "step": 8453
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.372632829440448e-08,
+      "loss": 0.5337,
+      "step": 8454
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3420592642151587e-08,
+      "loss": 0.842,
+      "step": 8455
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.311624676361258e-08,
+      "loss": 0.8071,
+      "step": 8456
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2813290701229074e-08,
+      "loss": 0.5475,
+      "step": 8457
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.251172449724616e-08,
+      "loss": 0.5784,
+      "step": 8458
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2211548193714634e-08,
+      "loss": 0.4639,
+      "step": 8459
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.1912761832491035e-08,
+      "loss": 0.5445,
+      "step": 8460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.161536545524202e-08,
+      "loss": 0.7351,
+      "step": 8461
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.131935910343442e-08,
+      "loss": 0.5612,
+      "step": 8462
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.102474281834633e-08,
+      "loss": 0.5088,
+      "step": 8463
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0731516641059335e-08,
+      "loss": 0.4758,
+      "step": 8464
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.043968061246183e-08,
+      "loss": 0.8004,
+      "step": 8465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.014923477325016e-08,
+      "loss": 0.6783,
+      "step": 8466
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9860179163921924e-08,
+      "loss": 0.6496,
+      "step": 8467
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9572513824785987e-08,
+      "loss": 0.6011,
+      "step": 8468
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9286238795955823e-08,
+      "loss": 0.5803,
+      "step": 8469
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.90013541173495e-08,
+      "loss": 0.9572,
+      "step": 8470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.871785982869191e-08,
+      "loss": 0.5314,
+      "step": 8471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8435755969515866e-08,
+      "loss": 0.5851,
+      "step": 8472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.815504257915769e-08,
+      "loss": 0.8586,
+      "step": 8473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7875719696762725e-08,
+      "loss": 0.6385,
+      "step": 8474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7597787361278718e-08,
+      "loss": 0.4967,
+      "step": 8475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7321245611462433e-08,
+      "loss": 0.6297,
+      "step": 8476
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.704609448587525e-08,
+      "loss": 0.5918,
+      "step": 8477
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.677233402288648e-08,
+      "loss": 0.7209,
+      "step": 8478
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.649996426066781e-08,
+      "loss": 0.5006,
+      "step": 8479
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6228985237202188e-08,
+      "loss": 0.6674,
+      "step": 8480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5959396990273834e-08,
+      "loss": 0.765,
+      "step": 8481
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.569119955747601e-08,
+      "loss": 0.7769,
+      "step": 8482
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.542439297620658e-08,
+      "loss": 0.5868,
+      "step": 8483
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5158977283671337e-08,
+      "loss": 0.6983,
+      "step": 8484
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4894952516878457e-08,
+      "loss": 0.5358,
+      "step": 8485
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4632318712646264e-08,
+      "loss": 0.5047,
+      "step": 8486
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4371075907596575e-08,
+      "loss": 0.5191,
+      "step": 8487
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.411122413815803e-08,
+      "loss": 0.6981,
+      "step": 8488
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3852763440566085e-08,
+      "loss": 0.8647,
+      "step": 8489
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.359569385085969e-08,
+      "loss": 0.6113,
+      "step": 8490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3340015404886838e-08,
+      "loss": 0.5416,
+      "step": 8491
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3085728138299012e-08,
+      "loss": 0.5881,
+      "step": 8492
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2832832086556735e-08,
+      "loss": 0.5404,
+      "step": 8493
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2581327284922905e-08,
+      "loss": 0.5435,
+      "step": 8494
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2331213768468363e-08,
+      "loss": 0.607,
+      "step": 8495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2082491572070763e-08,
+      "loss": 0.5295,
+      "step": 8496
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1835160730413472e-08,
+      "loss": 0.5039,
+      "step": 8497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.158922127798224e-08,
+      "loss": 0.7584,
+      "step": 8498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.134467324907519e-08,
+      "loss": 0.6996,
+      "step": 8499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.110151667778948e-08,
+      "loss": 0.6408,
+      "step": 8500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.085975159803466e-08,
+      "loss": 0.5642,
+      "step": 8501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0619378043521544e-08,
+      "loss": 0.6124,
+      "step": 8502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0380396047768868e-08,
+      "loss": 0.5956,
+      "step": 8503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0142805644102204e-08,
+      "loss": 0.5584,
+      "step": 8504
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9906606865650603e-08,
+      "loss": 0.805,
+      "step": 8505
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.967179974535105e-08,
+      "loss": 0.6321,
+      "step": 8506
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9438384315945137e-08,
+      "loss": 0.4637,
+      "step": 8507
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9206360609983487e-08,
+      "loss": 0.8445,
+      "step": 8508
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8975728659816894e-08,
+      "loss": 0.6075,
+      "step": 8509
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8746488497607406e-08,
+      "loss": 0.7298,
+      "step": 8510
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8518640155320567e-08,
+      "loss": 0.5895,
+      "step": 8511
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8292183664728734e-08,
+      "loss": 0.448,
+      "step": 8512
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8067119057409988e-08,
+      "loss": 0.9479,
+      "step": 8513
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7843446364746998e-08,
+      "loss": 0.5326,
+      "step": 8514
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7621165617929258e-08,
+      "loss": 0.5825,
+      "step": 8515
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.74002768479542e-08,
+      "loss": 0.5597,
+      "step": 8516
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.718078008562274e-08,
+      "loss": 0.7425,
+      "step": 8517
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6962675361540393e-08,
+      "loss": 0.5828,
+      "step": 8518
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.674596270612283e-08,
+      "loss": 0.6321,
+      "step": 8519
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6530642149588106e-08,
+      "loss": 0.5884,
+      "step": 8520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6316713721961086e-08,
+      "loss": 0.6428,
+      "step": 8521
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6104177453072355e-08,
+      "loss": 0.5841,
+      "step": 8522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.589303337256043e-08,
+      "loss": 0.7093,
+      "step": 8523
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.56832815098662e-08,
+      "loss": 0.7172,
+      "step": 8524
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5474921894238493e-08,
+      "loss": 0.5179,
+      "step": 8525
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5267954554731846e-08,
+      "loss": 0.7393,
+      "step": 8526
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5062379520206503e-08,
+      "loss": 0.9289,
+      "step": 8527
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.485819681932954e-08,
+      "loss": 0.8615,
+      "step": 8528
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.465540648057151e-08,
+      "loss": 0.527,
+      "step": 8529
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4454008532210905e-08,
+      "loss": 0.8953,
+      "step": 8530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4254003002330819e-08,
+      "loss": 0.5311,
+      "step": 8531
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.405538991882116e-08,
+      "loss": 0.5446,
+      "step": 8532
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3858169309376446e-08,
+      "loss": 0.7374,
+      "step": 8533
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3662341201498009e-08,
+      "loss": 0.7966,
+      "step": 8534
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3467905622494004e-08,
+      "loss": 0.8416,
+      "step": 8535
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.327486259947608e-08,
+      "loss": 0.6642,
+      "step": 8536
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3083212159363811e-08,
+      "loss": 0.5467,
+      "step": 8537
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2892954328880269e-08,
+      "loss": 0.46,
+      "step": 8538
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2704089134556452e-08,
+      "loss": 0.6724,
+      "step": 8539
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.251661660272685e-08,
+      "loss": 0.571,
+      "step": 8540
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2330536759536105e-08,
+      "loss": 0.5421,
+      "step": 8541
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.214584963093013e-08,
+      "loss": 0.6012,
+      "step": 8542
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1962555242661655e-08,
+      "loss": 0.5115,
+      "step": 8543
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.178065362029135e-08,
+      "loss": 0.795,
+      "step": 8544
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1600144789183366e-08,
+      "loss": 0.5711,
+      "step": 8545
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.142102877450868e-08,
+      "loss": 0.7052,
+      "step": 8546
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.124330560124287e-08,
+      "loss": 0.6368,
+      "step": 8547
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1066975294170557e-08,
+      "loss": 0.5875,
+      "step": 8548
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0892037877877627e-08,
+      "loss": 0.6193,
+      "step": 8549
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.071849337675901e-08,
+      "loss": 0.5142,
+      "step": 8550
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0546341815014238e-08,
+      "loss": 0.6516,
+      "step": 8551
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0375583216648554e-08,
+      "loss": 0.6884,
+      "step": 8552
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0206217605472912e-08,
+      "loss": 0.5168,
+      "step": 8553
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0038245005105085e-08,
+      "loss": 0.5648,
+      "step": 8554
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.871665438966338e-09,
+      "loss": 0.5865,
+      "step": 8555
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.706478930285868e-09,
+      "loss": 0.5856,
+      "step": 8556
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.542685502098581e-09,
+      "loss": 0.6174,
+      "step": 8557
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.380285177242875e-09,
+      "loss": 0.5537,
+      "step": 8558
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.219277978366192e-09,
+      "loss": 0.6287,
+      "step": 8559
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.05966392791835e-09,
+      "loss": 0.659,
+      "step": 8560
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.901443048157099e-09,
+      "loss": 0.5982,
+      "step": 8561
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.744615361144792e-09,
+      "loss": 0.715,
+      "step": 8562
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.589180888750603e-09,
+      "loss": 0.6056,
+      "step": 8563
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.435139652649416e-09,
+      "loss": 0.5578,
+      "step": 8564
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.282491674319604e-09,
+      "loss": 0.5485,
+      "step": 8565
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.131236975048584e-09,
+      "loss": 0.7312,
+      "step": 8566
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.981375575927264e-09,
+      "loss": 0.7878,
+      "step": 8567
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.83290749785226e-09,
+      "loss": 0.5771,
+      "step": 8568
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.685832761527012e-09,
+      "loss": 0.6566,
+      "step": 8569
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.54015138745956e-09,
+      "loss": 0.7201,
+      "step": 8570
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.395863395964764e-09,
+      "loss": 0.5784,
+      "step": 8571
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.252968807163197e-09,
+      "loss": 0.6764,
+      "step": 8572
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.11146764097892e-09,
+      "loss": 0.6178,
+      "step": 8573
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.971359917145038e-09,
+      "loss": 0.5333,
+      "step": 8574
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.832645655198145e-09,
+      "loss": 0.7885,
+      "step": 8575
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.695324874480547e-09,
+      "loss": 0.8194,
+      "step": 8576
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.5593975941402595e-09,
+      "loss": 0.5374,
+      "step": 8577
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.424863833133232e-09,
+      "loss": 0.6169,
+      "step": 8578
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.291723610216682e-09,
+      "loss": 0.7609,
+      "step": 8579
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.1599769439590896e-09,
+      "loss": 0.7286,
+      "step": 8580
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.029623852730204e-09,
+      "loss": 0.5457,
+      "step": 8581
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.900664354706598e-09,
+      "loss": 0.6196,
+      "step": 8582
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.773098467870553e-09,
+      "loss": 0.5192,
+      "step": 8583
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.646926210011172e-09,
+      "loss": 0.4802,
+      "step": 8584
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.522147598723271e-09,
+      "loss": 0.6534,
+      "step": 8585
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.398762651404044e-09,
+      "loss": 0.5516,
+      "step": 8586
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2767713852597265e-09,
+      "loss": 0.463,
+      "step": 8587
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.156173817302268e-09,
+      "loss": 0.8372,
+      "step": 8588
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.036969964347105e-09,
+      "loss": 0.5325,
+      "step": 8589
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.919159843015387e-09,
+      "loss": 0.7006,
+      "step": 8590
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.802743469737303e-09,
+      "loss": 0.5454,
+      "step": 8591
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.687720860745426e-09,
+      "loss": 0.8114,
+      "step": 8592
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.5740920320780365e-09,
+      "loss": 0.7573,
+      "step": 8593
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.461856999581349e-09,
+      "loss": 0.5645,
+      "step": 8594
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.351015778903955e-09,
+      "loss": 0.8898,
+      "step": 8595
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.241568385503492e-09,
+      "loss": 0.5763,
+      "step": 8596
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.133514834641083e-09,
+      "loss": 0.6513,
+      "step": 8597
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.026855141384678e-09,
+      "loss": 0.6691,
+      "step": 8598
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.921589320606822e-09,
+      "loss": 0.5235,
+      "step": 8599
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.8177173869857754e-09,
+      "loss": 0.515,
+      "step": 8600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7152393550055068e-09,
+      "loss": 0.9291,
+      "step": 8601
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6141552389568067e-09,
+      "loss": 0.676,
+      "step": 8602
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.514465052935068e-09,
+      "loss": 0.8124,
+      "step": 8603
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4161688108413917e-09,
+      "loss": 0.765,
+      "step": 8604
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.319266526382592e-09,
+      "loss": 0.6933,
+      "step": 8605
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.223758213071193e-09,
+      "loss": 0.5248,
+      "step": 8606
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1296438842243203e-09,
+      "loss": 0.5626,
+      "step": 8607
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0369235529659204e-09,
+      "loss": 0.6229,
+      "step": 8608
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.94559723222565e-09,
+      "loss": 0.4949,
+      "step": 8609
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8556649347377674e-09,
+      "loss": 0.572,
+      "step": 8610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.767126673043352e-09,
+      "loss": 0.8582,
+      "step": 8611
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.6799824594891943e-09,
+      "loss": 0.6172,
+      "step": 8612
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5942323062244646e-09,
+      "loss": 0.6145,
+      "step": 8613
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5098762252095952e-09,
+      "loss": 0.5129,
+      "step": 8614
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4269142282051792e-09,
+      "loss": 0.584,
+      "step": 8615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.345346326780851e-09,
+      "loss": 0.5876,
+      "step": 8616
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.265172532309734e-09,
+      "loss": 0.544,
+      "step": 8617
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.186392855971775e-09,
+      "loss": 0.6417,
+      "step": 8618
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.109007308753741e-09,
+      "loss": 0.5661,
+      "step": 8619
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0330159014447793e-09,
+      "loss": 0.703,
+      "step": 8620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9584186446419685e-09,
+      "loss": 0.6373,
+      "step": 8621
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.885215548748098e-09,
+      "loss": 0.7832,
+      "step": 8622
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8134066239705595e-09,
+      "loss": 0.6651,
+      "step": 8623
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.742991880321343e-09,
+      "loss": 0.7725,
+      "step": 8624
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6739713276203717e-09,
+      "loss": 0.6566,
+      "step": 8625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.606344975492169e-09,
+      "loss": 0.68,
+      "step": 8626
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.540112833366969e-09,
+      "loss": 0.6283,
+      "step": 8627
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4752749104796071e-09,
+      "loss": 0.6854,
+      "step": 8628
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4118312158717396e-09,
+      "loss": 0.8638,
+      "step": 8629
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3497817583907336e-09,
+      "loss": 0.6029,
+      "step": 8630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2891265466885572e-09,
+      "loss": 0.595,
+      "step": 8631
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2298655892217792e-09,
+      "loss": 0.5682,
+      "step": 8632
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1719988942571203e-09,
+      "loss": 0.8716,
+      "step": 8633
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1155264698603508e-09,
+      "loss": 0.5032,
+      "step": 8634
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0604483239085028e-09,
+      "loss": 0.5491,
+      "step": 8635
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0067644640809892e-09,
+      "loss": 0.6016,
+      "step": 8636
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.544748978629336e-10,
+      "loss": 0.7696,
+      "step": 8637
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.035796325473911e-10,
+      "loss": 0.8977,
+      "step": 8638
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.540786752297969e-10,
+      "loss": 0.5515,
+      "step": 8639
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.059720328135179e-10,
+      "loss": 0.542,
+      "step": 8640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.592597120065215e-10,
+      "loss": 0.6712,
+      "step": 8641
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.139417193235965e-10,
+      "loss": 0.7876,
+      "step": 8642
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.700180610819118e-10,
+      "loss": 0.9075,
+      "step": 8643
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.274887434087884e-10,
+      "loss": 0.552,
+      "step": 8644
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.863537722328172e-10,
+      "loss": 0.4719,
+      "step": 8645
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.466131532905206e-10,
+      "loss": 0.8249,
+      "step": 8646
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.082668921230216e-10,
+      "loss": 0.7595,
+      "step": 8647
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.713149940782646e-10,
+      "loss": 0.6249,
+      "step": 8648
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3575746430879473e-10,
+      "loss": 0.5928,
+      "step": 8649
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.0159430777286787e-10,
+      "loss": 0.7044,
+      "step": 8650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.6882552923334093e-10,
+      "loss": 0.6116,
+      "step": 8651
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3745113326100196e-10,
+      "loss": 0.8686,
+      "step": 8652
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.074711242301298e-10,
+      "loss": 0.6122,
+      "step": 8653
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.788855063218243e-10,
+      "loss": 0.6035,
+      "step": 8654
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5169428352067593e-10,
+      "loss": 0.6316,
+      "step": 8655
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.2589745962031673e-10,
+      "loss": 0.6147,
+      "step": 8656
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0149503821564887e-10,
+      "loss": 0.7718,
+      "step": 8657
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7848702271172636e-10,
+      "loss": 0.5479,
+      "step": 8658
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5687341631598352e-10,
+      "loss": 0.563,
+      "step": 8659
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3665422204156565e-10,
+      "loss": 0.5799,
+      "step": 8660
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1782944270843922e-10,
+      "loss": 0.4528,
+      "step": 8661
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0039908094228168e-10,
+      "loss": 0.6302,
+      "step": 8662
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.436313917226102e-11,
+      "loss": 0.5802,
+      "step": 8663
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.972161963547664e-11,
+      "loss": 0.5942,
+      "step": 8664
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.6474524373628656e-11,
+      "loss": 0.4988,
+      "step": 8665
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.462185523301799e-11,
+      "loss": 0.5514,
+      "step": 8666
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.416361386787692e-11,
+      "loss": 0.4907,
+      "step": 8667
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5099801735928297e-11,
+      "loss": 0.7608,
+      "step": 8668
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.7430420099495693e-11,
+      "loss": 0.5148,
+      "step": 8669
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.115547003105455e-11,
+      "loss": 0.8136,
+      "step": 8670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.2749524021299414e-12,
+      "loss": 0.8372,
+      "step": 8671
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.788867896619252e-12,
+      "loss": 0.6686,
+      "step": 8672
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.972169985797195e-13,
+      "loss": 0.5694,
+      "step": 8673
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.7263,
+      "step": 8674
+    },
+    {
+      "epoch": 2.0,
+      "step": 8674,
+      "total_flos": 357625141198848.0,
+      "train_loss": 0.7239196998150305,
+      "train_runtime": 87249.771,
+      "train_samples_per_second": 1.59,
+      "train_steps_per_second": 0.099
+    }
+  ],
+  "max_steps": 8674,
+  "num_train_epochs": 2,
+  "total_flos": 357625141198848.0,
+  "trial_name": null,
+  "trial_params": null
+}