diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14893 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 2478,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 1.9913,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.9995,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.2818,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.1715,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.0634,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.0393,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 1.0523,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 0.9802,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.9437,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.937,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 0.9299,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.8872,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 0.9059,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 0.8929,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8798,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 0.8417,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 0.8558,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.8442,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 0.7991,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.8078,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.8042,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 0.7828,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 0.7963,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.8151,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.767,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 0.8099,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.7901,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 0.7483,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 0.7507,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.7315,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 0.7198,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 0.7376,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.8e-06,
+      "loss": 0.7294,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.7233,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.7129,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.7133,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 0.7197,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 0.7276,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.04e-05,
+      "loss": 0.7226,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.7135,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 0.7342,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.6788,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 0.7131,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 0.6999,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6865,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 0.7051,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 0.7001,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.6801,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 0.6795,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7341,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.6975,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 0.6659,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 0.6925,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.7053,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.6726,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 0.6711,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.6671,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 0.699,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 0.6739,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6859,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 0.6727,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 0.6826,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.6624,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 0.6548,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.647,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.76e-05,
+      "loss": 0.6577,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 0.6559,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 0.6643,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.6558,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.6334,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 0.6625,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.6635,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 0.629,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 0.6623,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.6589,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999991454014607e-05,
+      "loss": 0.6421,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996581607303e-05,
+      "loss": 0.635,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999992308621909e-05,
+      "loss": 0.635,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999986326452582e-05,
+      "loss": 0.6433,
+      "step": 79
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999786351095475e-05,
+      "loss": 0.6407,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999692346059504e-05,
+      "loss": 0.6486,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999581249578582e-05,
+      "loss": 0.6556,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999453061842603e-05,
+      "loss": 0.6484,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999930778307066e-05,
+      "loss": 0.6187,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999914541351106e-05,
+      "loss": 0.645,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999896595344133e-05,
+      "loss": 0.6211,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998769403168195e-05,
+      "loss": 0.6208,
+      "step": 87
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999855576302761e-05,
+      "loss": 0.6302,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998325033384716e-05,
+      "loss": 0.6068,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998077214633884e-05,
+      "loss": 0.6485,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999781230719868e-05,
+      "loss": 0.6498,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999753031153189e-05,
+      "loss": 0.6245,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997231228115487e-05,
+      "loss": 0.6225,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996915057460675e-05,
+      "loss": 0.6147,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999658180010785e-05,
+      "loss": 0.6328,
+      "step": 95
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999623145662661e-05,
+      "loss": 0.6018,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995864027615765e-05,
+      "loss": 0.6226,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995479513703325e-05,
+      "loss": 0.6445,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995077915546496e-05,
+      "loss": 0.6056,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999465923383169e-05,
+      "loss": 0.6099,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994223469274517e-05,
+      "loss": 0.6124,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993770622619784e-05,
+      "loss": 0.5943,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993300694641495e-05,
+      "loss": 0.62,
+      "step": 103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999281368614285e-05,
+      "loss": 0.6059,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992309597956244e-05,
+      "loss": 0.5984,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999178843094326e-05,
+      "loss": 0.6013,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999125018599468e-05,
+      "loss": 0.5698,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999069486403046e-05,
+      "loss": 0.633,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990122465999767e-05,
+      "loss": 0.5813,
+      "step": 109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989532992880935e-05,
+      "loss": 0.6107,
+      "step": 110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988926445681495e-05,
+      "loss": 0.5881,
+      "step": 111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988302825438146e-05,
+      "loss": 0.6007,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998766213321679e-05,
+      "loss": 0.6006,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987004370112487e-05,
+      "loss": 0.6215,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986329537249493e-05,
+      "loss": 0.5807,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985637635781218e-05,
+      "loss": 0.6094,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998492866689027e-05,
+      "loss": 0.5676,
+      "step": 117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984202631788407e-05,
+      "loss": 0.5674,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983459531716575e-05,
+      "loss": 0.6145,
+      "step": 119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998269936794487e-05,
+      "loss": 0.6003,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981922141772566e-05,
+      "loss": 0.6236,
+      "step": 121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981127854528096e-05,
+      "loss": 0.5685,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998031650756905e-05,
+      "loss": 0.5974,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979488102282184e-05,
+      "loss": 0.5645,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978642640083408e-05,
+      "loss": 0.5997,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977780122417778e-05,
+      "loss": 0.5647,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976900550759506e-05,
+      "loss": 0.5767,
+      "step": 127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997600392661196e-05,
+      "loss": 0.5944,
+      "step": 128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975090251507637e-05,
+      "loss": 0.5918,
+      "step": 129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99741595270082e-05,
+      "loss": 0.5983,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973211754704432e-05,
+      "loss": 0.5516,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972246936216268e-05,
+      "loss": 0.6004,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971265073192768e-05,
+      "loss": 0.6102,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970266167312132e-05,
+      "loss": 0.6236,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969250220281687e-05,
+      "loss": 0.6037,
+      "step": 135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968217233837888e-05,
+      "loss": 0.6185,
+      "step": 136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967167209746307e-05,
+      "loss": 0.5885,
+      "step": 137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966100149801648e-05,
+      "loss": 0.5852,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965016055827725e-05,
+      "loss": 0.5845,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996391492967747e-05,
+      "loss": 0.5799,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996279677323292e-05,
+      "loss": 0.5856,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961661588405222e-05,
+      "loss": 0.5877,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996050937713464e-05,
+      "loss": 0.5774,
+      "step": 143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995934014139053e-05,
+      "loss": 0.5824,
+      "step": 144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995815388317134e-05,
+      "loss": 0.5633,
+      "step": 145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995695060450462e-05,
+      "loss": 0.5902,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955730307447015e-05,
+      "loss": 0.5802,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995449299408425e-05,
+      "loss": 0.5623,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995323866653113e-05,
+      "loss": 0.6402,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995196732693156e-05,
+      "loss": 0.607,
+      "step": 150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950678977458503e-05,
+      "loss": 0.5919,
+      "step": 151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949373620314007e-05,
+      "loss": 0.5848,
+      "step": 152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948051257729184e-05,
+      "loss": 0.5825,
+      "step": 153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946711891964206e-05,
+      "loss": 0.5832,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994535552530832e-05,
+      "loss": 0.5459,
+      "step": 155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943982160079823e-05,
+      "loss": 0.5808,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994259179862606e-05,
+      "loss": 0.619,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941184443323442e-05,
+      "loss": 0.5743,
+      "step": 158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939760096577408e-05,
+      "loss": 0.5853,
+      "step": 159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938318760822457e-05,
+      "loss": 0.5632,
+      "step": 160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993686043852211e-05,
+      "loss": 0.5485,
+      "step": 161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993538513216892e-05,
+      "loss": 0.5555,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933892844284495e-05,
+      "loss": 0.5746,
+      "step": 163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.5472,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930857334153374e-05,
+      "loss": 0.5795,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929314117094967e-05,
+      "loss": 0.5647,
+      "step": 166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927753928881874e-05,
+      "loss": 0.5504,
+      "step": 167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992617677218077e-05,
+      "loss": 0.5638,
+      "step": 168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992458264968732e-05,
+      "loss": 0.5719,
+      "step": 169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922971564126195e-05,
+      "loss": 0.5577,
+      "step": 170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921343518251057e-05,
+      "loss": 0.5816,
+      "step": 171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991969851484456e-05,
+      "loss": 0.5554,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918036556718337e-05,
+      "loss": 0.5564,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916357646713006e-05,
+      "loss": 0.5628,
+      "step": 174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914661787698148e-05,
+      "loss": 0.5728,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912948982572325e-05,
+      "loss": 0.5513,
+      "step": 176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991121923426306e-05,
+      "loss": 0.5641,
+      "step": 177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909472545726827e-05,
+      "loss": 0.5734,
+      "step": 178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907708919949068e-05,
+      "loss": 0.5649,
+      "step": 179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990592835994416e-05,
+      "loss": 0.5526,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990413086875544e-05,
+      "loss": 0.5588,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902316449455168e-05,
+      "loss": 0.5233,
+      "step": 182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900485105144544e-05,
+      "loss": 0.5623,
+      "step": 183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98986368389537e-05,
+      "loss": 0.575,
+      "step": 184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896771654041687e-05,
+      "loss": 0.5678,
+      "step": 185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989488955359647e-05,
+      "loss": 0.5628,
+      "step": 186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892990540834935e-05,
+      "loss": 0.5452,
+      "step": 187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989107461900286e-05,
+      "loss": 0.5203,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889141791374942e-05,
+      "loss": 0.5631,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988719206125476e-05,
+      "loss": 0.5536,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988522543197479e-05,
+      "loss": 0.536,
+      "step": 191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988324190689639e-05,
+      "loss": 0.5711,
+      "step": 192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881241489409786e-05,
+      "loss": 0.5401,
+      "step": 193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879224182934095e-05,
+      "loss": 0.5516,
+      "step": 194
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987718999091729e-05,
+      "loss": 0.5823,
+      "step": 195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98751389168362e-05,
+      "loss": 0.5299,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873070964196522e-05,
+      "loss": 0.5503,
+      "step": 197
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870986136532792e-05,
+      "loss": 0.5579,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986888443740839e-05,
+      "loss": 0.5616,
+      "step": 199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866765870415538e-05,
+      "loss": 0.5614,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864630439175282e-05,
+      "loss": 0.537,
+      "step": 201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862478147337493e-05,
+      "loss": 0.5636,
+      "step": 202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860308998580862e-05,
+      "loss": 0.5409,
+      "step": 203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858122996612897e-05,
+      "loss": 0.5445,
+      "step": 204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98559201451699e-05,
+      "loss": 0.5659,
+      "step": 205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853700448016982e-05,
+      "loss": 0.5566,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985146390894804e-05,
+      "loss": 0.5513,
+      "step": 207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849210531785763e-05,
+      "loss": 0.5427,
+      "step": 208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846940320381614e-05,
+      "loss": 0.5142,
+      "step": 209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844653278615836e-05,
+      "loss": 0.5659,
+      "step": 210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984234941039743e-05,
+      "loss": 0.555,
+      "step": 211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984002871966416e-05,
+      "loss": 0.548,
+      "step": 212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837691210382547e-05,
+      "loss": 0.5665,
+      "step": 213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835336886547852e-05,
+      "loss": 0.5517,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832965752184084e-05,
+      "loss": 0.5703,
+      "step": 215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830577811343973e-05,
+      "loss": 0.5617,
+      "step": 216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982817306810898e-05,
+      "loss": 0.5451,
+      "step": 217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825751526589292e-05,
+      "loss": 0.5557,
+      "step": 218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9823313190923797e-05,
+      "loss": 0.5539,
+      "step": 219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820858065280088e-05,
+      "loss": 0.5477,
+      "step": 220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981838615385446e-05,
+      "loss": 0.5268,
+      "step": 221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98158974608719e-05,
+      "loss": 0.5311,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813391990586073e-05,
+      "loss": 0.5201,
+      "step": 223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810869747279323e-05,
+      "loss": 0.5023,
+      "step": 224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808330735262657e-05,
+      "loss": 0.5448,
+      "step": 225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980577495887575e-05,
+      "loss": 0.5372,
+      "step": 226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803202422486925e-05,
+      "loss": 0.5521,
+      "step": 227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800613130493158e-05,
+      "loss": 0.5377,
+      "step": 228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798007087320052e-05,
+      "loss": 0.5216,
+      "step": 229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795384297421857e-05,
+      "loss": 0.5447,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792744765281435e-05,
+      "loss": 0.5286,
+      "step": 231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790088495410265e-05,
+      "loss": 0.5542,
+      "step": 232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787415492348435e-05,
+      "loss": 0.5131,
+      "step": 233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784725760664632e-05,
+      "loss": 0.549,
+      "step": 234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782019304956144e-05,
+      "loss": 0.5513,
+      "step": 235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779296129848834e-05,
+      "loss": 0.528,
+      "step": 236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776556239997146e-05,
+      "loss": 0.5389,
+      "step": 237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773799640084087e-05,
+      "loss": 0.5136,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771026334821237e-05,
+      "loss": 0.512,
+      "step": 239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768236328948718e-05,
+      "loss": 0.5144,
+      "step": 240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765429627235194e-05,
+      "loss": 0.5197,
+      "step": 241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762606234477882e-05,
+      "loss": 0.5507,
+      "step": 242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975976615550251e-05,
+      "loss": 0.5426,
+      "step": 243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975690939516333e-05,
+      "loss": 0.5412,
+      "step": 244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975403595834312e-05,
+      "loss": 0.5451,
+      "step": 245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751145849953135e-05,
+      "loss": 0.5651,
+      "step": 246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974823907493315e-05,
+      "loss": 0.533,
+      "step": 247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974531563825141e-05,
+      "loss": 0.5056,
+      "step": 248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742375544904654e-05,
+      "loss": 0.5397,
+      "step": 249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739418799918073e-05,
+      "loss": 0.5451,
+      "step": 250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736445408345327e-05,
+      "loss": 0.5315,
+      "step": 251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9733455375268532e-05,
+      "loss": 0.5363,
+      "step": 252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.5292,
+      "step": 253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727425405073444e-05,
+      "loss": 0.5285,
+      "step": 254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972438547826156e-05,
+      "loss": 0.5304,
+      "step": 255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721328930558423e-05,
+      "loss": 0.5135,
+      "step": 256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718255767188275e-05,
+      "loss": 0.5314,
+      "step": 257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715165993403756e-05,
+      "loss": 0.526,
+      "step": 258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712059614485898e-05,
+      "loss": 0.4925,
+      "step": 259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708936635744118e-05,
+      "loss": 0.5075,
+      "step": 260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705797062516204e-05,
+      "loss": 0.5341,
+      "step": 261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702640900168296e-05,
+      "loss": 0.5187,
+      "step": 262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969946815409491e-05,
+      "loss": 0.5595,
+      "step": 263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696278829718882e-05,
+      "loss": 0.528,
+      "step": 264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693072932491405e-05,
+      "loss": 0.5372,
+      "step": 265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689850467891987e-05,
+      "loss": 0.5081,
+      "step": 266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686611441428455e-05,
+      "loss": 0.5354,
+      "step": 267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968335585863694e-05,
+      "loss": 0.5245,
+      "step": 268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968008372508187e-05,
+      "loss": 0.5303,
+      "step": 269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676795046355986e-05,
+      "loss": 0.5392,
+      "step": 270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967348982808027e-05,
+      "loss": 0.5431,
+      "step": 271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670168075903997e-05,
+      "loss": 0.5327,
+      "step": 272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666829795504693e-05,
+      "loss": 0.4999,
+      "step": 273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663474992588144e-05,
+      "loss": 0.5324,
+      "step": 274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660103672888367e-05,
+      "loss": 0.5718,
+      "step": 275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656715842167606e-05,
+      "loss": 0.5277,
+      "step": 276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9653311506216336e-05,
+      "loss": 0.5268,
+      "step": 277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964989067085324e-05,
+      "loss": 0.4909,
+      "step": 278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646453341925195e-05,
+      "loss": 0.5031,
+      "step": 279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642999525307278e-05,
+      "loss": 0.5209,
+      "step": 280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963952922690274e-05,
+      "loss": 0.5505,
+      "step": 281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636042452643004e-05,
+      "loss": 0.5124,
+      "step": 282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632539208487655e-05,
+      "loss": 0.5431,
+      "step": 283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962901950042443e-05,
+      "loss": 0.5196,
+      "step": 284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9625483334469198e-05,
+      "loss": 0.5508,
+      "step": 285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621930716665972e-05,
+      "loss": 0.4924,
+      "step": 286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961836165308687e-05,
+      "loss": 0.5237,
+      "step": 287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614776149832127e-05,
+      "loss": 0.5179,
+      "step": 288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611174213030072e-05,
+      "loss": 0.528,
+      "step": 289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960755584883713e-05,
+      "loss": 0.4924,
+      "step": 290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603921063437795e-05,
+      "loss": 0.5236,
+      "step": 291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600269863044633e-05,
+      "loss": 0.4999,
+      "step": 292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959660225389826e-05,
+      "loss": 0.515,
+      "step": 293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959291824226735e-05,
+      "loss": 0.5225,
+      "step": 294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95892178344486e-05,
+      "loss": 0.5352,
+      "step": 295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958550103676674e-05,
+      "loss": 0.5207,
+      "step": 296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581767855574508e-05,
+      "loss": 0.5381,
+      "step": 297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578018297252647e-05,
+      "loss": 0.5017,
+      "step": 298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574252368209886e-05,
+      "loss": 0.545,
+      "step": 299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570470074882947e-05,
+      "loss": 0.5213,
+      "step": 300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566671423736517e-05,
+      "loss": 0.4945,
+      "step": 301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956285642126323e-05,
+      "loss": 0.5312,
+      "step": 302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559025073983678e-05,
+      "loss": 0.5142,
+      "step": 303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95551773884464e-05,
+      "loss": 0.5047,
+      "step": 304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551313371227835e-05,
+      "loss": 0.5079,
+      "step": 305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547433028932357e-05,
+      "loss": 0.4976,
+      "step": 306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543536368192234e-05,
+      "loss": 0.5101,
+      "step": 307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953962339566763e-05,
+      "loss": 0.5298,
+      "step": 308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535694118046584e-05,
+      "loss": 0.5078,
+      "step": 309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9531748542045005e-05,
+      "loss": 0.4886,
+      "step": 310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952778667440666e-05,
+      "loss": 0.5129,
+      "step": 311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523808521903165e-05,
+      "loss": 0.5504,
+      "step": 312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951981409133396e-05,
+      "loss": 0.5343,
+      "step": 313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951580338952632e-05,
+      "loss": 0.4974,
+      "step": 314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511776423335327e-05,
+      "loss": 0.5297,
+      "step": 315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507733199643853e-05,
+      "loss": 0.4978,
+      "step": 316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503673725362567e-05,
+      "loss": 0.4955,
+      "step": 317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949959800742991e-05,
+      "loss": 0.4958,
+      "step": 318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495506052812087e-05,
+      "loss": 0.5252,
+      "step": 319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491397868503057e-05,
+      "loss": 0.5035,
+      "step": 320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487273461524517e-05,
+      "loss": 0.4894,
+      "step": 321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483132838925884e-05,
+      "loss": 0.5151,
+      "step": 322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947897600778431e-05,
+      "loss": 0.5125,
+      "step": 323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947480297520463e-05,
+      "loss": 0.4965,
+      "step": 324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470613748319377e-05,
+      "loss": 0.4979,
+      "step": 325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466408334288775e-05,
+      "loss": 0.5293,
+      "step": 326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462186740300697e-05,
+      "loss": 0.511,
+      "step": 327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457948973570684e-05,
+      "loss": 0.508,
+      "step": 328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453695041341912e-05,
+      "loss": 0.4945,
+      "step": 329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449424950885193e-05,
+      "loss": 0.5285,
+      "step": 330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944513870949895e-05,
+      "loss": 0.4986,
+      "step": 331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944083632450921e-05,
+      "loss": 0.5139,
+      "step": 332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436517803269603e-05,
+      "loss": 0.5044,
+      "step": 333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943218315316133e-05,
+      "loss": 0.4864,
+      "step": 334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942783238159317e-05,
+      "loss": 0.5083,
+      "step": 335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942346549600144e-05,
+      "loss": 0.4963,
+      "step": 336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419082503850006e-05,
+      "loss": 0.4845,
+      "step": 337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414683412630277e-05,
+      "loss": 0.5195,
+      "step": 338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941026822986116e-05,
+      "loss": 0.5067,
+      "step": 339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405836963089066e-05,
+      "loss": 0.5078,
+      "step": 340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401389619887913e-05,
+      "loss": 0.4856,
+      "step": 341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.5059,
+      "step": 342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939244673463143e-05,
+      "loss": 0.5345,
+      "step": 343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387951207861256e-05,
+      "loss": 0.4904,
+      "step": 344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383439635232296e-05,
+      "loss": 0.5258,
+      "step": 345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378912024455726e-05,
+      "loss": 0.5126,
+      "step": 346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374368383270114e-05,
+      "loss": 0.5072,
+      "step": 347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369808719441448e-05,
+      "loss": 0.4864,
+      "step": 348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936523304076309e-05,
+      "loss": 0.5021,
+      "step": 349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360641355055772e-05,
+      "loss": 0.518,
+      "step": 350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356033670167595e-05,
+      "loss": 0.5124,
+      "step": 351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351409993973997e-05,
+      "loss": 0.498,
+      "step": 352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346770334377756e-05,
+      "loss": 0.4948,
+      "step": 353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342114699308962e-05,
+      "loss": 0.5154,
+      "step": 354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933744309672501e-05,
+      "loss": 0.5183,
+      "step": 355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9332755534610594e-05,
+      "loss": 0.5128,
+      "step": 356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932805202097768e-05,
+      "loss": 0.5021,
+      "step": 357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323332563865502e-05,
+      "loss": 0.5105,
+      "step": 358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318597171340535e-05,
+      "loss": 0.5145,
+      "step": 359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313845851496507e-05,
+      "loss": 0.5005,
+      "step": 360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309078612454356e-05,
+      "loss": 0.4978,
+      "step": 361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304295462362232e-05,
+      "loss": 0.5167,
+      "step": 362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299496409395482e-05,
+      "loss": 0.4906,
+      "step": 363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929468146175663e-05,
+      "loss": 0.4838,
+      "step": 364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289850627675378e-05,
+      "loss": 0.5025,
+      "step": 365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285003915408575e-05,
+      "loss": 0.4933,
+      "step": 366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928014133324019e-05,
+      "loss": 0.528,
+      "step": 367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275262889481357e-05,
+      "loss": 0.5087,
+      "step": 368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927036859247028e-05,
+      "loss": 0.5236,
+      "step": 369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265458450572287e-05,
+      "loss": 0.4816,
+      "step": 370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260532472179777e-05,
+      "loss": 0.5024,
+      "step": 371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255590665712214e-05,
+      "loss": 0.4682,
+      "step": 372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925063303961612e-05,
+      "loss": 0.4821,
+      "step": 373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924565960236506e-05,
+      "loss": 0.5146,
+      "step": 374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924067036245961e-05,
+      "loss": 0.5209,
+      "step": 375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9235665328427373e-05,
+      "loss": 0.4922,
+      "step": 376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230644508822936e-05,
+      "loss": 0.4815,
+      "step": 377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225607912227864e-05,
+      "loss": 0.5039,
+      "step": 378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220555547250698e-05,
+      "loss": 0.4814,
+      "step": 379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215487422526926e-05,
+      "loss": 0.5071,
+      "step": 380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921040354671897e-05,
+      "loss": 0.485,
+      "step": 381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205303928516174e-05,
+      "loss": 0.4765,
+      "step": 382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200188576634796e-05,
+      "loss": 0.5019,
+      "step": 383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919505749981798e-05,
+      "loss": 0.4924,
+      "step": 384
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189910706835738e-05,
+      "loss": 0.5159,
+      "step": 385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184748206484963e-05,
+      "loss": 0.4948,
+      "step": 386
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179570007589384e-05,
+      "loss": 0.47,
+      "step": 387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917437611899956e-05,
+      "loss": 0.4971,
+      "step": 388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169166549592876e-05,
+      "loss": 0.4922,
+      "step": 389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163941308273504e-05,
+      "loss": 0.5081,
+      "step": 390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915870040397242e-05,
+      "loss": 0.4825,
+      "step": 391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153443845647357e-05,
+      "loss": 0.479,
+      "step": 392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148171642282812e-05,
+      "loss": 0.4911,
+      "step": 393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142883802890018e-05,
+      "loss": 0.475,
+      "step": 394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137580336506935e-05,
+      "loss": 0.4859,
+      "step": 395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132261252198236e-05,
+      "loss": 0.5185,
+      "step": 396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126926559055277e-05,
+      "loss": 0.5015,
+      "step": 397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121576266196103e-05,
+      "loss": 0.496,
+      "step": 398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911621038276542e-05,
+      "loss": 0.4812,
+      "step": 399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110828917934582e-05,
+      "loss": 0.5053,
+      "step": 400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910543188090157e-05,
+      "loss": 0.4853,
+      "step": 401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100019280890984e-05,
+      "loss": 0.4885,
+      "step": 402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094591127154025e-05,
+      "loss": 0.4964,
+      "step": 403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089147428968478e-05,
+      "loss": 0.502,
+      "step": 404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083688195638694e-05,
+      "loss": 0.4986,
+      "step": 405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078213436495582e-05,
+      "loss": 0.4785,
+      "step": 406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907272316089658e-05,
+      "loss": 0.4792,
+      "step": 407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067217378225655e-05,
+      "loss": 0.4838,
+      "step": 408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9061696097893275e-05,
+      "loss": 0.5066,
+      "step": 409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056159329336395e-05,
+      "loss": 0.4995,
+      "step": 410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050607082018437e-05,
+      "loss": 0.4752,
+      "step": 411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045039365429297e-05,
+      "loss": 0.4771,
+      "step": 412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039456189085292e-05,
+      "loss": 0.4803,
+      "step": 413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033857562529176e-05,
+      "loss": 0.4933,
+      "step": 414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90282434953301e-05,
+      "loss": 0.4511,
+      "step": 415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902261399708362e-05,
+      "loss": 0.5109,
+      "step": 416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016969077411645e-05,
+      "loss": 0.5039,
+      "step": 417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011308745962466e-05,
+      "loss": 0.4824,
+      "step": 418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90056330124107e-05,
+      "loss": 0.4861,
+      "step": 419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999941886457292e-05,
+      "loss": 0.5236,
+      "step": 420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994235377829506e-05,
+      "loss": 0.4924,
+      "step": 421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988513496280885e-05,
+      "loss": 0.4543,
+      "step": 422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982776251591247e-05,
+      "loss": 0.5049,
+      "step": 423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977023653566683e-05,
+      "loss": 0.4977,
+      "step": 424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971255712039513e-05,
+      "loss": 0.4889,
+      "step": 425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965472436868288e-05,
+      "loss": 0.4774,
+      "step": 426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895967383793776e-05,
+      "loss": 0.5017,
+      "step": 427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953859925158878e-05,
+      "loss": 0.4707,
+      "step": 428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894803070846877e-05,
+      "loss": 0.4558,
+      "step": 429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8942186197830713e-05,
+      "loss": 0.4975,
+      "step": 430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.4721,
+      "step": 431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893045133469455e-05,
+      "loss": 0.4858,
+      "step": 432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924561002253645e-05,
+      "loss": 0.4677,
+      "step": 433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8918655415979143e-05,
+      "loss": 0.4585,
+      "step": 434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891273458596486e-05,
+      "loss": 0.4967,
+      "step": 435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890679852233065e-05,
+      "loss": 0.4899,
+      "step": 436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900847235222427e-05,
+      "loss": 0.4607,
+      "step": 437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894880734812106e-05,
+      "loss": 0.4808,
+      "step": 438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888899031297622e-05,
+      "loss": 0.5129,
+      "step": 439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882902134902876e-05,
+      "loss": 0.4647,
+      "step": 440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876890055877745e-05,
+      "loss": 0.4623,
+      "step": 441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8870862804498065e-05,
+      "loss": 0.482,
+      "step": 442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864820391065592e-05,
+      "loss": 0.4611,
+      "step": 443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858762825908e-05,
+      "loss": 0.4921,
+      "step": 444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885269011937886e-05,
+      "loss": 0.4523,
+      "step": 445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846602281857632e-05,
+      "loss": 0.4942,
+      "step": 446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840499323749624e-05,
+      "loss": 0.4512,
+      "step": 447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834381255485996e-05,
+      "loss": 0.4762,
+      "step": 448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882824808752373e-05,
+      "loss": 0.4408,
+      "step": 449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882209983034562e-05,
+      "loss": 0.4847,
+      "step": 450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8815936494460252e-05,
+      "loss": 0.4791,
+      "step": 451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809758090401982e-05,
+      "loss": 0.4945,
+      "step": 452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803564628730916e-05,
+      "loss": 0.4574,
+      "step": 453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.87973561200329e-05,
+      "loss": 0.4815,
+      "step": 454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791132574919504e-05,
+      "loss": 0.4942,
+      "step": 455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878489400402799e-05,
+      "loss": 0.4907,
+      "step": 456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8778640418021306e-05,
+      "loss": 0.4849,
+      "step": 457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877237182758806e-05,
+      "loss": 0.5018,
+      "step": 458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766088243442514e-05,
+      "loss": 0.47,
+      "step": 459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759789676324546e-05,
+      "loss": 0.4459,
+      "step": 460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8753476136999655e-05,
+      "loss": 0.4594,
+      "step": 461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874714763625892e-05,
+      "loss": 0.5154,
+      "step": 462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874080418491899e-05,
+      "loss": 0.4688,
+      "step": 463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734445793822088e-05,
+      "loss": 0.4841,
+      "step": 464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728072473835944e-05,
+      "loss": 0.5177,
+      "step": 465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721684235853824e-05,
+      "loss": 0.4939,
+      "step": 466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8715281090794483e-05,
+      "loss": 0.4978,
+      "step": 467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708863049602163e-05,
+      "loss": 0.46,
+      "step": 468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702430123246554e-05,
+      "loss": 0.4615,
+      "step": 469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695982322722805e-05,
+      "loss": 0.4762,
+      "step": 470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689519659051467e-05,
+      "loss": 0.4798,
+      "step": 471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868304214327851e-05,
+      "loss": 0.4643,
+      "step": 472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676549786475283e-05,
+      "loss": 0.4823,
+      "step": 473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867004259973851e-05,
+      "loss": 0.4724,
+      "step": 474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8663520594190255e-05,
+      "loss": 0.466,
+      "step": 475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86569837809779e-05,
+      "loss": 0.4677,
+      "step": 476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865043217127416e-05,
+      "loss": 0.5098,
+      "step": 477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643865776277016e-05,
+      "loss": 0.4739,
+      "step": 478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863728460720974e-05,
+      "loss": 0.5054,
+      "step": 479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630688675320844e-05,
+      "loss": 0.4888,
+      "step": 480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624077991884072e-05,
+      "loss": 0.4924,
+      "step": 481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861745256819839e-05,
+      "loss": 0.4801,
+      "step": 482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610812415587948e-05,
+      "loss": 0.4513,
+      "step": 483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604157545402085e-05,
+      "loss": 0.4794,
+      "step": 484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859748796901527e-05,
+      "loss": 0.4541,
+      "step": 485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590803697827138e-05,
+      "loss": 0.4823,
+      "step": 486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584104743262417e-05,
+      "loss": 0.4803,
+      "step": 487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577391116770944e-05,
+      "loss": 0.4854,
+      "step": 488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570662829827632e-05,
+      "loss": 0.4892,
+      "step": 489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563919893932446e-05,
+      "loss": 0.4587,
+      "step": 490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557162320610392e-05,
+      "loss": 0.4725,
+      "step": 491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550390121411497e-05,
+      "loss": 0.446,
+      "step": 492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854360330791078e-05,
+      "loss": 0.4885,
+      "step": 493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536801891708248e-05,
+      "loss": 0.4714,
+      "step": 494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529985884428855e-05,
+      "loss": 0.4717,
+      "step": 495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852315529772251e-05,
+      "loss": 0.4821,
+      "step": 496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516310143264027e-05,
+      "loss": 0.4767,
+      "step": 497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509450432753123e-05,
+      "loss": 0.4868,
+      "step": 498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502576177914394e-05,
+      "loss": 0.4921,
+      "step": 499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84956873904973e-05,
+      "loss": 0.4592,
+      "step": 500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488784082276137e-05,
+      "loss": 0.4542,
+      "step": 501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481866265050013e-05,
+      "loss": 0.4772,
+      "step": 502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474933950642846e-05,
+      "loss": 0.4715,
+      "step": 503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467987150903325e-05,
+      "loss": 0.4998,
+      "step": 504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461025877704905e-05,
+      "loss": 0.4722,
+      "step": 505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454050142945768e-05,
+      "loss": 0.4799,
+      "step": 506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447059958548822e-05,
+      "loss": 0.4849,
+      "step": 507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.844005533646167e-05,
+      "loss": 0.4646,
+      "step": 508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843303628865659e-05,
+      "loss": 0.4933,
+      "step": 509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426002827130517e-05,
+      "loss": 0.4846,
+      "step": 510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418954963905027e-05,
+      "loss": 0.4754,
+      "step": 511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411892711026307e-05,
+      "loss": 0.479,
+      "step": 512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404816080565133e-05,
+      "loss": 0.4646,
+      "step": 513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397725084616867e-05,
+      "loss": 0.4697,
+      "step": 514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839061973530142e-05,
+      "loss": 0.4606,
+      "step": 515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383500044763226e-05,
+      "loss": 0.4852,
+      "step": 516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8376366025171247e-05,
+      "loss": 0.4649,
+      "step": 517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369217688718922e-05,
+      "loss": 0.4941,
+      "step": 518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362055047624175e-05,
+      "loss": 0.468,
+      "step": 519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.4597,
+      "step": 520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347686900501288e-05,
+      "loss": 0.474,
+      "step": 521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340481419031146e-05,
+      "loss": 0.4687,
+      "step": 522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333261682034523e-05,
+      "loss": 0.4604,
+      "step": 523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8326027701851376e-05,
+      "loss": 0.4756,
+      "step": 524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318779490846005e-05,
+      "loss": 0.435,
+      "step": 525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311517061407025e-05,
+      "loss": 0.483,
+      "step": 526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304240425947363e-05,
+      "loss": 0.4997,
+      "step": 527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829694959690422e-05,
+      "loss": 0.4834,
+      "step": 528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289644586739068e-05,
+      "loss": 0.4857,
+      "step": 529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8282325407937594e-05,
+      "loss": 0.4434,
+      "step": 530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274992073009736e-05,
+      "loss": 0.4476,
+      "step": 531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267644594489588e-05,
+      "loss": 0.4615,
+      "step": 532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8260282984935457e-05,
+      "loss": 0.4187,
+      "step": 533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252907256929777e-05,
+      "loss": 0.4593,
+      "step": 534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245517423079117e-05,
+      "loss": 0.455,
+      "step": 535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238113496014166e-05,
+      "loss": 0.4872,
+      "step": 536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8230695488389688e-05,
+      "loss": 0.4391,
+      "step": 537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223263412884527e-05,
+      "loss": 0.4579,
+      "step": 538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215817282201565e-05,
+      "loss": 0.468,
+      "step": 539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82083571090677e-05,
+      "loss": 0.4856,
+      "step": 540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820088290623384e-05,
+      "loss": 0.4546,
+      "step": 541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819339468647487e-05,
+      "loss": 0.4764,
+      "step": 542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818589246258964e-05,
+      "loss": 0.455,
+      "step": 543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178376247400922e-05,
+      "loss": 0.4445,
+      "step": 544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170846053755413e-05,
+      "loss": 0.4643,
+      "step": 545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163301894523695e-05,
+      "loss": 0.462,
+      "step": 546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155743782600226e-05,
+      "loss": 0.428,
+      "step": 547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8148171730903305e-05,
+      "loss": 0.4661,
+      "step": 548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8140585752375063e-05,
+      "loss": 0.4804,
+      "step": 549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132985859981435e-05,
+      "loss": 0.4677,
+      "step": 550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812537206671213e-05,
+      "loss": 0.4914,
+      "step": 551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117744385580627e-05,
+      "loss": 0.4659,
+      "step": 552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811010282962413e-05,
+      "loss": 0.4455,
+      "step": 553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102447411903567e-05,
+      "loss": 0.4627,
+      "step": 554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8094778145503555e-05,
+      "loss": 0.4329,
+      "step": 555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087095043532382e-05,
+      "loss": 0.4405,
+      "step": 556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079398119121983e-05,
+      "loss": 0.4691,
+      "step": 557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8071687385427922e-05,
+      "loss": 0.4666,
+      "step": 558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8063962855629357e-05,
+      "loss": 0.4531,
+      "step": 559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056224542929035e-05,
+      "loss": 0.4642,
+      "step": 560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804847246055326e-05,
+      "loss": 0.4747,
+      "step": 561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804070662175186e-05,
+      "loss": 0.4563,
+      "step": 562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8032927039798194e-05,
+      "loss": 0.4572,
+      "step": 563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025133727989095e-05,
+      "loss": 0.4562,
+      "step": 564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801732669964487e-05,
+      "loss": 0.4538,
+      "step": 565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800950596810927e-05,
+      "loss": 0.4614,
+      "step": 566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001671546749466e-05,
+      "loss": 0.447,
+      "step": 567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7993823448956027e-05,
+      "loss": 0.4616,
+      "step": 568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985961688142902e-05,
+      "loss": 0.4722,
+      "step": 569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797808627774738e-05,
+      "loss": 0.4658,
+      "step": 570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970197231230103e-05,
+      "loss": 0.4391,
+      "step": 571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7962294562074997e-05,
+      "loss": 0.4354,
+      "step": 572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795437828378929e-05,
+      "loss": 0.4241,
+      "step": 573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794644840990345e-05,
+      "loss": 0.462,
+      "step": 574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7938504953971198e-05,
+      "loss": 0.4574,
+      "step": 575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793054792956947e-05,
+      "loss": 0.4517,
+      "step": 576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792257735029839e-05,
+      "loss": 0.4618,
+      "step": 577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791459322978124e-05,
+      "loss": 0.4512,
+      "step": 578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906595581664462e-05,
+      "loss": 0.4567,
+      "step": 579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898584419617616e-05,
+      "loss": 0.4643,
+      "step": 580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890559757333346e-05,
+      "loss": 0.4717,
+      "step": 581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882521608527393e-05,
+      "loss": 0.4833,
+      "step": 582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874469986938526e-05,
+      "loss": 0.4578,
+      "step": 583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866404906328564e-05,
+      "loss": 0.4441,
+      "step": 584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858326380482313e-05,
+      "loss": 0.4698,
+      "step": 585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850234423207568e-05,
+      "loss": 0.4644,
+      "step": 586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784212904833508e-05,
+      "loss": 0.4457,
+      "step": 587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834010269718526e-05,
+      "loss": 0.4524,
+      "step": 588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825878101234506e-05,
+      "loss": 0.4414,
+      "step": 589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7817732556782493e-05,
+      "loss": 0.4449,
+      "step": 590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780957365028483e-05,
+      "loss": 0.4413,
+      "step": 591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801401395686696e-05,
+      "loss": 0.4581,
+      "step": 592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793215806956087e-05,
+      "loss": 0.4772,
+      "step": 593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785016898083786e-05,
+      "loss": 0.4645,
+      "step": 594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777680468308334e-05,
+      "loss": 0.4484,
+      "step": 595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768579175991048e-05,
+      "loss": 0.4514,
+      "step": 596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776034039086592e-05,
+      "loss": 0.4553,
+      "step": 597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752088341789666e-05,
+      "loss": 0.4484,
+      "step": 598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7743823042866664e-05,
+      "loss": 0.4739,
+      "step": 599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735544508223933e-05,
+      "loss": 0.4721,
+      "step": 600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727252752011125e-05,
+      "loss": 0.4329,
+      "step": 601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718947788400492e-05,
+      "loss": 0.4399,
+      "step": 602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771062963158684e-05,
+      "loss": 0.4502,
+      "step": 603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702298295787547e-05,
+      "loss": 0.4342,
+      "step": 604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693953795242505e-05,
+      "loss": 0.4661,
+      "step": 605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768559614421411e-05,
+      "loss": 0.4357,
+      "step": 606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767722535698724e-05,
+      "loss": 0.4719,
+      "step": 607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766884144786921e-05,
+      "loss": 0.4592,
+      "step": 608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.478,
+      "step": 609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652034321301108e-05,
+      "loss": 0.4712,
+      "step": 610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643611132577724e-05,
+      "loss": 0.4171,
+      "step": 611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763517487941652e-05,
+      "loss": 0.4222,
+      "step": 612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626725576236718e-05,
+      "loss": 0.4542,
+      "step": 613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618263237479835e-05,
+      "loss": 0.4364,
+      "step": 614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609787877609678e-05,
+      "loss": 0.4518,
+      "step": 615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7601299511112312e-05,
+      "loss": 0.4355,
+      "step": 616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759279815249603e-05,
+      "loss": 0.4314,
+      "step": 617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758428381629132e-05,
+      "loss": 0.4407,
+      "step": 618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575756517050864e-05,
+      "loss": 0.4214,
+      "step": 619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75672162693495e-05,
+      "loss": 0.4648,
+      "step": 620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558663087784195e-05,
+      "loss": 0.442,
+      "step": 621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550096986974017e-05,
+      "loss": 0.4255,
+      "step": 622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541517981560127e-05,
+      "loss": 0.4492,
+      "step": 623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753292608620573e-05,
+      "loss": 0.4345,
+      "step": 624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7524321315596072e-05,
+      "loss": 0.4658,
+      "step": 625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515703684438397e-05,
+      "loss": 0.4582,
+      "step": 626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750707320746194e-05,
+      "loss": 0.4611,
+      "step": 627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498429899417887e-05,
+      "loss": 0.4444,
+      "step": 628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489773775079354e-05,
+      "loss": 0.4521,
+      "step": 629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481104849241357e-05,
+      "loss": 0.4537,
+      "step": 630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747242313672081e-05,
+      "loss": 0.4465,
+      "step": 631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7463728652356465e-05,
+      "loss": 0.4323,
+      "step": 632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455021411008906e-05,
+      "loss": 0.4609,
+      "step": 633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446301427560533e-05,
+      "loss": 0.437,
+      "step": 634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437568716915506e-05,
+      "loss": 0.4551,
+      "step": 635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428823293999757e-05,
+      "loss": 0.4605,
+      "step": 636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7420065173760936e-05,
+      "loss": 0.4404,
+      "step": 637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741129437116839e-05,
+      "loss": 0.4449,
+      "step": 638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402510901213158e-05,
+      "loss": 0.4355,
+      "step": 639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7393714778907914e-05,
+      "loss": 0.4533,
+      "step": 640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7384906019286964e-05,
+      "loss": 0.4544,
+      "step": 641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7376084637406222e-05,
+      "loss": 0.4274,
+      "step": 642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367250648343167e-05,
+      "loss": 0.4484,
+      "step": 643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358404067196822e-05,
+      "loss": 0.464,
+      "step": 644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349544909087737e-05,
+      "loss": 0.4339,
+      "step": 645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340673189157962e-05,
+      "loss": 0.472,
+      "step": 646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331788922571016e-05,
+      "loss": 0.4664,
+      "step": 647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7322892124511862e-05,
+      "loss": 0.4415,
+      "step": 648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313982810186876e-05,
+      "loss": 0.4403,
+      "step": 649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730506099482384e-05,
+      "loss": 0.4498,
+      "step": 650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7296126693671886e-05,
+      "loss": 0.429,
+      "step": 651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72871799220015e-05,
+      "loss": 0.4147,
+      "step": 652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727822069510448e-05,
+      "loss": 0.4405,
+      "step": 653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7269249028293907e-05,
+      "loss": 0.4689,
+      "step": 654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726026493690413e-05,
+      "loss": 0.4523,
+      "step": 655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251268436290728e-05,
+      "loss": 0.4388,
+      "step": 656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7242259541830497e-05,
+      "loss": 0.4238,
+      "step": 657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233238268921414e-05,
+      "loss": 0.4692,
+      "step": 658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722420463298261e-05,
+      "loss": 0.4474,
+      "step": 659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721515864945435e-05,
+      "loss": 0.4615,
+      "step": 660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206100333798003e-05,
+      "loss": 0.4512,
+      "step": 661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719702970149601e-05,
+      "loss": 0.4683,
+      "step": 662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718794676805188e-05,
+      "loss": 0.4582,
+      "step": 663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717885154899013e-05,
+      "loss": 0.4505,
+      "step": 664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716974405985628e-05,
+      "loss": 0.4477,
+      "step": 665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160624316216825e-05,
+      "loss": 0.4717,
+      "step": 666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715149233365921e-05,
+      "loss": 0.4593,
+      "step": 667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142348127791784e-05,
+      "loss": 0.4652,
+      "step": 668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133191714243805e-05,
+      "loss": 0.4262,
+      "step": 669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124023108665383e-05,
+      "loss": 0.4533,
+      "step": 670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711484232672747e-05,
+      "loss": 0.4346,
+      "step": 671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710564938412184e-05,
+      "loss": 0.4677,
+      "step": 672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7096444296561036e-05,
+      "loss": 0.4264,
+      "step": 673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087227079778373e-05,
+      "loss": 0.4469,
+      "step": 674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7077997749527884e-05,
+      "loss": 0.4459,
+      "step": 675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706875632158432e-05,
+      "loss": 0.4708,
+      "step": 676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7059502811743095e-05,
+      "loss": 0.4502,
+      "step": 677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050237235820287e-05,
+      "loss": 0.4014,
+      "step": 678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040959609652593e-05,
+      "loss": 0.429,
+      "step": 679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031669949097297e-05,
+      "loss": 0.4504,
+      "step": 680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022368270032268e-05,
+      "loss": 0.4217,
+      "step": 681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7013054588355904e-05,
+      "loss": 0.4335,
+      "step": 682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700372891998712e-05,
+      "loss": 0.4425,
+      "step": 683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994391280865327e-05,
+      "loss": 0.4135,
+      "step": 684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985041686950388e-05,
+      "loss": 0.4439,
+      "step": 685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69756801542226e-05,
+      "loss": 0.4339,
+      "step": 686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966306698682672e-05,
+      "loss": 0.4282,
+      "step": 687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956921336351682e-05,
+      "loss": 0.4663,
+      "step": 688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947524083271063e-05,
+      "loss": 0.4353,
+      "step": 689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693811495550258e-05,
+      "loss": 0.4468,
+      "step": 690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6928693969128275e-05,
+      "loss": 0.4402,
+      "step": 691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691926114025048e-05,
+      "loss": 0.4229,
+      "step": 692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690981648499176e-05,
+      "loss": 0.4248,
+      "step": 693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900360019494882e-05,
+      "loss": 0.4593,
+      "step": 694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890891759922822e-05,
+      "loss": 0.4262,
+      "step": 695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6881411722458688e-05,
+      "loss": 0.4763,
+      "step": 696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871919923305742e-05,
+      "loss": 0.4548,
+      "step": 697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.4048,
+      "step": 698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852901104846902e-05,
+      "loss": 0.4546,
+      "step": 699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843374118047917e-05,
+      "loss": 0.4326,
+      "step": 700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683383543457388e-05,
+      "loss": 0.445,
+      "step": 701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6824285070728278e-05,
+      "loss": 0.4492,
+      "step": 702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681472304283457e-05,
+      "loss": 0.4293,
+      "step": 703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6805149367236145e-05,
+      "loss": 0.4275,
+      "step": 704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795564060296295e-05,
+      "loss": 0.4112,
+      "step": 705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785967138398205e-05,
+      "loss": 0.457,
+      "step": 706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776358617944907e-05,
+      "loss": 0.438,
+      "step": 707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676673851535925e-05,
+      "loss": 0.437,
+      "step": 708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675710684708389e-05,
+      "loss": 0.4348,
+      "step": 709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6747463629581247e-05,
+      "loss": 0.4539,
+      "step": 710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6737808879333477e-05,
+      "loss": 0.434,
+      "step": 711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6728142612842447e-05,
+      "loss": 0.4525,
+      "step": 712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718464846629726e-05,
+      "loss": 0.4161,
+      "step": 713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708775597236507e-05,
+      "loss": 0.4359,
+      "step": 714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6699074881223635e-05,
+      "loss": 0.4355,
+      "step": 715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6689362715171546e-05,
+      "loss": 0.4288,
+      "step": 716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679639115680247e-05,
+      "loss": 0.4215,
+      "step": 717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666990409936928e-05,
+      "loss": 0.441,
+      "step": 718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6660157682877707e-05,
+      "loss": 0.4479,
+      "step": 719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665039988286408e-05,
+      "loss": 0.4305,
+      "step": 720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66406307160064e-05,
+      "loss": 0.4647,
+      "step": 721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630850199002097e-05,
+      "loss": 0.4612,
+      "step": 722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6621058348568008e-05,
+      "loss": 0.4414,
+      "step": 723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6611255181440324e-05,
+      "loss": 0.4221,
+      "step": 724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6601440714374597e-05,
+      "loss": 0.4086,
+      "step": 725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591614964145685e-05,
+      "loss": 0.4209,
+      "step": 726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658177794754773e-05,
+      "loss": 0.4401,
+      "step": 727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6571929681394136e-05,
+      "loss": 0.4554,
+      "step": 728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6562070182517524e-05,
+      "loss": 0.4454,
+      "step": 729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552199467769728e-05,
+      "loss": 0.43,
+      "step": 730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6542317554021736e-05,
+      "loss": 0.4534,
+      "step": 731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532424458163692e-05,
+      "loss": 0.4146,
+      "step": 732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6522520197104853e-05,
+      "loss": 0.4327,
+      "step": 733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651260478777354e-05,
+      "loss": 0.4319,
+      "step": 734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502678247117146e-05,
+      "loss": 0.4556,
+      "step": 735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649274059210209e-05,
+      "loss": 0.4471,
+      "step": 736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648279183971378e-05,
+      "loss": 0.4395,
+      "step": 737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647283200695659e-05,
+      "loss": 0.4723,
+      "step": 738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6462861110853846e-05,
+      "loss": 0.4461,
+      "step": 739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645287916844777e-05,
+      "loss": 0.4227,
+      "step": 740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442886196799465e-05,
+      "loss": 0.4436,
+      "step": 741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432882212988895e-05,
+      "loss": 0.4349,
+      "step": 742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642286723411484e-05,
+      "loss": 0.4043,
+      "step": 743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412841277294865e-05,
+      "loss": 0.4025,
+      "step": 744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6402804359665313e-05,
+      "loss": 0.4471,
+      "step": 745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392756498381253e-05,
+      "loss": 0.4317,
+      "step": 746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382697710616458e-05,
+      "loss": 0.4158,
+      "step": 747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372628013563386e-05,
+      "loss": 0.4383,
+      "step": 748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362547424433122e-05,
+      "loss": 0.4402,
+      "step": 749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352455960455385e-05,
+      "loss": 0.465,
+      "step": 750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634235363887848e-05,
+      "loss": 0.4291,
+      "step": 751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332240476969262e-05,
+      "loss": 0.434,
+      "step": 752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6322116492013116e-05,
+      "loss": 0.4744,
+      "step": 753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631198170131393e-05,
+      "loss": 0.4285,
+      "step": 754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630183612219406e-05,
+      "loss": 0.3978,
+      "step": 755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6291679771994293e-05,
+      "loss": 0.4456,
+      "step": 756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628151266807384e-05,
+      "loss": 0.4536,
+      "step": 757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271334827810288e-05,
+      "loss": 0.4297,
+      "step": 758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6261146268599564e-05,
+      "loss": 0.4326,
+      "step": 759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250947007855925e-05,
+      "loss": 0.4446,
+      "step": 760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240737063011922e-05,
+      "loss": 0.4159,
+      "step": 761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623051645151836e-05,
+      "loss": 0.4589,
+      "step": 762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622028519084428e-05,
+      "loss": 0.4542,
+      "step": 763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6210043298476918e-05,
+      "loss": 0.4389,
+      "step": 764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199790791921693e-05,
+      "loss": 0.4158,
+      "step": 765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189527688702156e-05,
+      "loss": 0.4381,
+      "step": 766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179254006359974e-05,
+      "loss": 0.4322,
+      "step": 767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168969762454897e-05,
+      "loss": 0.4092,
+      "step": 768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6158674974564715e-05,
+      "loss": 0.4512,
+      "step": 769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614836966028526e-05,
+      "loss": 0.4482,
+      "step": 770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6138053837230345e-05,
+      "loss": 0.4305,
+      "step": 771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6127727523031737e-05,
+      "loss": 0.425,
+      "step": 772
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117390735339146e-05,
+      "loss": 0.4307,
+      "step": 773
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610704349182018e-05,
+      "loss": 0.4273,
+      "step": 774
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609668581016032e-05,
+      "loss": 0.461,
+      "step": 775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608631770806288e-05,
+      "loss": 0.4128,
+      "step": 776
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607593920324899e-05,
+      "loss": 0.441,
+      "step": 777
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6065550313457567e-05,
+      "loss": 0.4147,
+      "step": 778
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605515105644526e-05,
+      "loss": 0.446,
+      "step": 779
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044741449986458e-05,
+      "loss": 0.416,
+      "step": 780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603432151187323e-05,
+      "loss": 0.4059,
+      "step": 781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6023891259915295e-05,
+      "loss": 0.4447,
+      "step": 782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6013450711940017e-05,
+      "loss": 0.4213,
+      "step": 783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002999885792348e-05,
+      "loss": 0.4068,
+      "step": 784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992538799334808e-05,
+      "loss": 0.4339,
+      "step": 785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598206747044746e-05,
+      "loss": 0.4222,
+      "step": 786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.4304,
+      "step": 787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596109415699106e-05,
+      "loss": 0.3919,
+      "step": 788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5950592208269536e-05,
+      "loss": 0.4146,
+      "step": 789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940080088813196e-05,
+      "loss": 0.417,
+      "step": 790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929557816589312e-05,
+      "loss": 0.433,
+      "step": 791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5919025409582537e-05,
+      "loss": 0.4548,
+      "step": 792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590848288579482e-05,
+      "loss": 0.4339,
+      "step": 793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589793026324541e-05,
+      "loss": 0.4313,
+      "step": 794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5887367559970825e-05,
+      "loss": 0.4384,
+      "step": 795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5876794794024803e-05,
+      "loss": 0.4083,
+      "step": 796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5866211983478287e-05,
+      "loss": 0.4103,
+      "step": 797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855619146419382e-05,
+      "loss": 0.4274,
+      "step": 798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584501630095334e-05,
+      "loss": 0.3996,
+      "step": 799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583440346520251e-05,
+      "loss": 0.4102,
+      "step": 800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5823780657306313e-05,
+      "loss": 0.4245,
+      "step": 801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5813147895421234e-05,
+      "loss": 0.4447,
+      "step": 802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802505197720753e-05,
+      "loss": 0.4191,
+      "step": 803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791852582395334e-05,
+      "loss": 0.4045,
+      "step": 804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5781190067652406e-05,
+      "loss": 0.4404,
+      "step": 805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5770517671716298e-05,
+      "loss": 0.4146,
+      "step": 806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575983541282824e-05,
+      "loss": 0.4356,
+      "step": 807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574914330924632e-05,
+      "loss": 0.4397,
+      "step": 808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573844137924545e-05,
+      "loss": 0.4431,
+      "step": 809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727729641117338e-05,
+      "loss": 0.4259,
+      "step": 810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5717008113170453e-05,
+      "loss": 0.3869,
+      "step": 811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570627681373e-05,
+      "loss": 0.4319,
+      "step": 812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569553576113789e-05,
+      "loss": 0.4293,
+      "step": 813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568478497375269e-05,
+      "loss": 0.4248,
+      "step": 814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567402446994962e-05,
+      "loss": 0.4184,
+      "step": 815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663254268120497e-05,
+      "loss": 0.417,
+      "step": 816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652474386673727e-05,
+      "loss": 0.4232,
+      "step": 817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641684844034243e-05,
+      "loss": 0.4256,
+      "step": 818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630885658643508e-05,
+      "loss": 0.4195,
+      "step": 819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5620076848959447e-05,
+      "loss": 0.4307,
+      "step": 820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5609258433456454e-05,
+      "loss": 0.4308,
+      "step": 821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5598430430625335e-05,
+      "loss": 0.3929,
+      "step": 822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5587592858973274e-05,
+      "loss": 0.4479,
+      "step": 823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557674573702382e-05,
+      "loss": 0.4125,
+      "step": 824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565889083316847e-05,
+      "loss": 0.4459,
+      "step": 825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5555022916408506e-05,
+      "loss": 0.411,
+      "step": 826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5544147254871226e-05,
+      "loss": 0.3612,
+      "step": 827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553326211729365e-05,
+      "loss": 0.3307,
+      "step": 828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552236752228063e-05,
+      "loss": 0.3249,
+      "step": 829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5511463488453168e-05,
+      "loss": 0.3467,
+      "step": 830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5500550034448415e-05,
+      "loss": 0.3438,
+      "step": 831
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489627178919612e-05,
+      "loss": 0.367,
+      "step": 832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5478694940536062e-05,
+      "loss": 0.3561,
+      "step": 833
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546775333798313e-05,
+      "loss": 0.3251,
+      "step": 834
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545680238996216e-05,
+      "loss": 0.3738,
+      "step": 835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544584211519049e-05,
+      "loss": 0.3421,
+      "step": 836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543487253240138e-05,
+      "loss": 0.3502,
+      "step": 837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5423893660344017e-05,
+      "loss": 0.3564,
+      "step": 838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541290551778345e-05,
+      "loss": 0.3338,
+      "step": 839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540190812350059e-05,
+      "loss": 0.3445,
+      "step": 840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539090149629214e-05,
+      "loss": 0.3403,
+      "step": 841
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5379885654970595e-05,
+      "loss": 0.3348,
+      "step": 842
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536886061836421e-05,
+      "loss": 0.3496,
+      "step": 843
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357826405316936e-05,
+      "loss": 0.317,
+      "step": 844
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346783034688422e-05,
+      "loss": 0.3317,
+      "step": 845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5335730525353962e-05,
+      "loss": 0.3349,
+      "step": 846
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5324668896204475e-05,
+      "loss": 0.344,
+      "step": 847
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313598166146464e-05,
+      "loss": 0.3625,
+      "step": 848
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302518354101992e-05,
+      "loss": 0.3379,
+      "step": 849
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291429479008634e-05,
+      "loss": 0.3496,
+      "step": 850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528033155981947e-05,
+      "loss": 0.3437,
+      "step": 851
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5269224615503025e-05,
+      "loss": 0.3323,
+      "step": 852
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5258108665043262e-05,
+      "loss": 0.347,
+      "step": 853
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246983727439529e-05,
+      "loss": 0.3499,
+      "step": 854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5235849821706531e-05,
+      "loss": 0.3406,
+      "step": 855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5224706966874316e-05,
+      "loss": 0.3796,
+      "step": 856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213555181988213e-05,
+      "loss": 0.3478,
+      "step": 857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202394486108823e-05,
+      "loss": 0.3377,
+      "step": 858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5191224898311973e-05,
+      "loss": 0.3327,
+      "step": 859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180046437688693e-05,
+      "loss": 0.3369,
+      "step": 860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5168859123345172e-05,
+      "loss": 0.365,
+      "step": 861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157662974402737e-05,
+      "loss": 0.3239,
+      "step": 862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514645800999781e-05,
+      "loss": 0.3325,
+      "step": 863
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135244249281884e-05,
+      "loss": 0.356,
+      "step": 864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512402171142149e-05,
+      "loss": 0.3311,
+      "step": 865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112790415598155e-05,
+      "loss": 0.3395,
+      "step": 866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101550381008377e-05,
+      "loss": 0.3469,
+      "step": 867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5090301626863588e-05,
+      "loss": 0.3596,
+      "step": 868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079044172390127e-05,
+      "loss": 0.3399,
+      "step": 869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5067778036829204e-05,
+      "loss": 0.35,
+      "step": 870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056503239436862e-05,
+      "loss": 0.3542,
+      "step": 871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5045219799483955e-05,
+      "loss": 0.3359,
+      "step": 872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5033927736256107e-05,
+      "loss": 0.3314,
+      "step": 873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022627069053673e-05,
+      "loss": 0.3291,
+      "step": 874
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5011317817191724e-05,
+      "loss": 0.3304,
+      "step": 875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3354,
+      "step": 876
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988673636822884e-05,
+      "loss": 0.3272,
+      "step": 877
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497733874701936e-05,
+      "loss": 0.3427,
+      "step": 878
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965995349962987e-05,
+      "loss": 0.3595,
+      "step": 879
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.495464346504187e-05,
+      "loss": 0.3411,
+      "step": 880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4943283111658614e-05,
+      "loss": 0.3403,
+      "step": 881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.49319143092303e-05,
+      "loss": 0.3389,
+      "step": 882
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492053707718846e-05,
+      "loss": 0.3326,
+      "step": 883
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909151434979019e-05,
+      "loss": 0.3466,
+      "step": 884
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897757402062285e-05,
+      "loss": 0.342,
+      "step": 885
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886354997912906e-05,
+      "loss": 0.3435,
+      "step": 886
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4874944242019838e-05,
+      "loss": 0.327,
+      "step": 887
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4863525153886314e-05,
+      "loss": 0.3095,
+      "step": 888
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4852097753029803e-05,
+      "loss": 0.3542,
+      "step": 889
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840662058981983e-05,
+      "loss": 0.3395,
+      "step": 890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829218091288713e-05,
+      "loss": 0.3157,
+      "step": 891
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481776586950999e-05,
+      "loss": 0.3724,
+      "step": 892
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806305413219915e-05,
+      "loss": 0.3542,
+      "step": 893
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4794836742006667e-05,
+      "loss": 0.3423,
+      "step": 894
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4783359875472462e-05,
+      "loss": 0.3431,
+      "step": 895
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771874833233534e-05,
+      "loss": 0.3152,
+      "step": 896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476038163492008e-05,
+      "loss": 0.3485,
+      "step": 897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4748880300176241e-05,
+      "loss": 0.3372,
+      "step": 898
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4737370848660062e-05,
+      "loss": 0.3541,
+      "step": 899
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725853300043472e-05,
+      "loss": 0.3584,
+      "step": 900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714327674012224e-05,
+      "loss": 0.3064,
+      "step": 901
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4702793990265882e-05,
+      "loss": 0.3387,
+      "step": 902
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691252268517794e-05,
+      "loss": 0.334,
+      "step": 903
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4679702528495032e-05,
+      "loss": 0.3454,
+      "step": 904
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668144789938376e-05,
+      "loss": 0.3189,
+      "step": 905
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656579072602281e-05,
+      "loss": 0.3353,
+      "step": 906
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645005396254842e-05,
+      "loss": 0.3269,
+      "step": 907
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463342378067774e-05,
+      "loss": 0.3204,
+      "step": 908
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621834245666254e-05,
+      "loss": 0.3368,
+      "step": 909
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4610236811029171e-05,
+      "loss": 0.3371,
+      "step": 910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598631496588799e-05,
+      "loss": 0.3325,
+      "step": 911
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587018322180906e-05,
+      "loss": 0.3291,
+      "step": 912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4575397307654697e-05,
+      "loss": 0.3207,
+      "step": 913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563768472872775e-05,
+      "loss": 0.3225,
+      "step": 914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552131837711108e-05,
+      "loss": 0.3342,
+      "step": 915
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4540487422059001e-05,
+      "loss": 0.3453,
+      "step": 916
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452883524581906e-05,
+      "loss": 0.3494,
+      "step": 917
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4517175328907141e-05,
+      "loss": 0.3463,
+      "step": 918
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505507691252346e-05,
+      "loss": 0.3473,
+      "step": 919
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4493832352796969e-05,
+      "loss": 0.3498,
+      "step": 920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4482149333496455e-05,
+      "loss": 0.3397,
+      "step": 921
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4470458653319397e-05,
+      "loss": 0.3262,
+      "step": 922
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458760332247468e-05,
+      "loss": 0.3482,
+      "step": 923
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4447054390275401e-05,
+      "loss": 0.3339,
+      "step": 924
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4435340847410961e-05,
+      "loss": 0.3508,
+      "step": 925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423619723674901e-05,
+      "loss": 0.332,
+      "step": 926
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411891039100934e-05,
+      "loss": 0.3482,
+      "step": 927
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400154813735685e-05,
+      "loss": 0.3579,
+      "step": 928
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388411067638685e-05,
+      "loss": 0.3411,
+      "step": 929
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4376659820882308e-05,
+      "loss": 0.3185,
+      "step": 930
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364901093551748e-05,
+      "loss": 0.3322,
+      "step": 931
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435313490574499e-05,
+      "loss": 0.3372,
+      "step": 932
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4341361277572766e-05,
+      "loss": 0.3214,
+      "step": 933
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329580229158526e-05,
+      "loss": 0.3326,
+      "step": 934
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317791780638406e-05,
+      "loss": 0.34,
+      "step": 935
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305995952161189e-05,
+      "loss": 0.3496,
+      "step": 936
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4294192763888269e-05,
+      "loss": 0.3271,
+      "step": 937
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.428238223599362e-05,
+      "loss": 0.3476,
+      "step": 938
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4270564388663761e-05,
+      "loss": 0.3286,
+      "step": 939
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258739242097725e-05,
+      "loss": 0.3414,
+      "step": 940
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246906816507015e-05,
+      "loss": 0.3236,
+      "step": 941
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235067132115581e-05,
+      "loss": 0.3155,
+      "step": 942
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4223220209159778e-05,
+      "loss": 0.3447,
+      "step": 943
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211366067888326e-05,
+      "loss": 0.3216,
+      "step": 944
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4199504728562294e-05,
+      "loss": 0.3522,
+      "step": 945
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4187636211455044e-05,
+      "loss": 0.3266,
+      "step": 946
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417576053685222e-05,
+      "loss": 0.3491,
+      "step": 947
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4163877725051677e-05,
+      "loss": 0.3458,
+      "step": 948
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4151987796363496e-05,
+      "loss": 0.3393,
+      "step": 949
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4140090771109896e-05,
+      "loss": 0.3594,
+      "step": 950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4128186669625247e-05,
+      "loss": 0.3539,
+      "step": 951
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116275512256001e-05,
+      "loss": 0.3176,
+      "step": 952
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104357319360672e-05,
+      "loss": 0.3794,
+      "step": 953
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092432111309804e-05,
+      "loss": 0.3255,
+      "step": 954
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080499908485924e-05,
+      "loss": 0.3351,
+      "step": 955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4068560731283518e-05,
+      "loss": 0.3415,
+      "step": 956
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4056614600108998e-05,
+      "loss": 0.3456,
+      "step": 957
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044661535380652e-05,
+      "loss": 0.3424,
+      "step": 958
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032701557528622e-05,
+      "loss": 0.3496,
+      "step": 959
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020734686994875e-05,
+      "loss": 0.3496,
+      "step": 960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008760944233142e-05,
+      "loss": 0.3405,
+      "step": 961
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3996780349708914e-05,
+      "loss": 0.3579,
+      "step": 962
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3984792923899387e-05,
+      "loss": 0.3421,
+      "step": 963
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972798687293434e-05,
+      "loss": 0.3351,
+      "step": 964
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.327,
+      "step": 965
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948789863705914e-05,
+      "loss": 0.3351,
+      "step": 966
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3936775317760158e-05,
+      "loss": 0.3525,
+      "step": 967
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392475404308953e-05,
+      "loss": 0.335,
+      "step": 968
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912726060240754e-05,
+      "loss": 0.3327,
+      "step": 969
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900691389772025e-05,
+      "loss": 0.3272,
+      "step": 970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3888650052252969e-05,
+      "loss": 0.3655,
+      "step": 971
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38766020682646e-05,
+      "loss": 0.3417,
+      "step": 972
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3864547458399302e-05,
+      "loss": 0.3221,
+      "step": 973
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852486243260774e-05,
+      "loss": 0.34,
+      "step": 974
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840418443464015e-05,
+      "loss": 0.3566,
+      "step": 975
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382834407963527e-05,
+      "loss": 0.339,
+      "step": 976
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3816263172412007e-05,
+      "loss": 0.3606,
+      "step": 977
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804175742442878e-05,
+      "loss": 0.3481,
+      "step": 978
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.379208181038768e-05,
+      "loss": 0.3251,
+      "step": 979
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3779981396917332e-05,
+      "loss": 0.3213,
+      "step": 980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376787452271382e-05,
+      "loss": 0.3276,
+      "step": 981
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375576120847018e-05,
+      "loss": 0.3395,
+      "step": 982
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3743641474890454e-05,
+      "loss": 0.3397,
+      "step": 983
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731515342689654e-05,
+      "loss": 0.3642,
+      "step": 984
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719382832593728e-05,
+      "loss": 0.3572,
+      "step": 985
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707243965339532e-05,
+      "loss": 0.3513,
+      "step": 986
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3695098761674779e-05,
+      "loss": 0.3314,
+      "step": 987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682947242358013e-05,
+      "loss": 0.3328,
+      "step": 988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3670789428158585e-05,
+      "loss": 0.3472,
+      "step": 989
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3658625339856586e-05,
+      "loss": 0.3574,
+      "step": 990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3646454998242841e-05,
+      "loss": 0.3318,
+      "step": 991
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3634278424118867e-05,
+      "loss": 0.3189,
+      "step": 992
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622095638296827e-05,
+      "loss": 0.3197,
+      "step": 993
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609906661599501e-05,
+      "loss": 0.3228,
+      "step": 994
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3597711514860255e-05,
+      "loss": 0.3437,
+      "step": 995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3585510218922997e-05,
+      "loss": 0.3307,
+      "step": 996
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573302794642147e-05,
+      "loss": 0.3265,
+      "step": 997
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561089262882594e-05,
+      "loss": 0.3366,
+      "step": 998
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3548869644519677e-05,
+      "loss": 0.3314,
+      "step": 999
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353664396043913e-05,
+      "loss": 0.338,
+      "step": 1000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352441223153706e-05,
+      "loss": 0.3345,
+      "step": 1001
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512174478719896e-05,
+      "loss": 0.3375,
+      "step": 1002
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499930722904373e-05,
+      "loss": 0.3393,
+      "step": 1003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3487680985017479e-05,
+      "loss": 0.3326,
+      "step": 1004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3475425285996438e-05,
+      "loss": 0.3422,
+      "step": 1005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463163646788648e-05,
+      "loss": 0.3432,
+      "step": 1006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345089608835167e-05,
+      "loss": 0.3497,
+      "step": 1007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438622631653178e-05,
+      "loss": 0.3622,
+      "step": 1008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3426343297670927e-05,
+      "loss": 0.349,
+      "step": 1009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341405810739272e-05,
+      "loss": 0.3131,
+      "step": 1010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340176708181637e-05,
+      "loss": 0.3257,
+      "step": 1011
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3389470241949658e-05,
+      "loss": 0.3165,
+      "step": 1012
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377167608810314e-05,
+      "loss": 0.3208,
+      "step": 1013
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364859203425953e-05,
+      "loss": 0.3483,
+      "step": 1014
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352545046834076e-05,
+      "loss": 0.336,
+      "step": 1015
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340225160081995e-05,
+      "loss": 0.3141,
+      "step": 1016
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327899564226826e-05,
+      "loss": 0.3652,
+      "step": 1017
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315568280335445e-05,
+      "loss": 0.3256,
+      "step": 1018
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3303231329484442e-05,
+      "loss": 0.3129,
+      "step": 1019
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.32908887327601e-05,
+      "loss": 0.3045,
+      "step": 1020
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278540511258351e-05,
+      "loss": 0.3316,
+      "step": 1021
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266186686084736e-05,
+      "loss": 0.3089,
+      "step": 1022
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3253827278354378e-05,
+      "loss": 0.3537,
+      "step": 1023
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241462309191942e-05,
+      "loss": 0.3179,
+      "step": 1024
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3229091799731592e-05,
+      "loss": 0.3539,
+      "step": 1025
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.321671577111697e-05,
+      "loss": 0.3381,
+      "step": 1026
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320433424450115e-05,
+      "loss": 0.3369,
+      "step": 1027
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191947241046603e-05,
+      "loss": 0.3355,
+      "step": 1028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317955478192515e-05,
+      "loss": 0.3258,
+      "step": 1029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167156888317956e-05,
+      "loss": 0.3335,
+      "step": 1030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3154753581415458e-05,
+      "loss": 0.3274,
+      "step": 1031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3142344882417355e-05,
+      "loss": 0.3318,
+      "step": 1032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3129930812532557e-05,
+      "loss": 0.3513,
+      "step": 1033
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3117511392979157e-05,
+      "loss": 0.3245,
+      "step": 1034
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310508664498439e-05,
+      "loss": 0.3113,
+      "step": 1035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.30926565897846e-05,
+      "loss": 0.3412,
+      "step": 1036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3080221248625202e-05,
+      "loss": 0.3282,
+      "step": 1037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306778064276064e-05,
+      "loss": 0.3337,
+      "step": 1038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055334793454364e-05,
+      "loss": 0.3114,
+      "step": 1039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304288372197879e-05,
+      "loss": 0.3572,
+      "step": 1040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030427449615241e-05,
+      "loss": 0.3281,
+      "step": 1041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301796599765395e-05,
+      "loss": 0.3438,
+      "step": 1042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005499387393991e-05,
+      "loss": 0.3472,
+      "step": 1043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299302764014326e-05,
+      "loss": 0.3212,
+      "step": 1044
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2980550777218426e-05,
+      "loss": 0.3272,
+      "step": 1045
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2968068819944914e-05,
+      "loss": 0.3497,
+      "step": 1046
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2955581789656844e-05,
+      "loss": 0.3427,
+      "step": 1047
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943089707697008e-05,
+      "loss": 0.349,
+      "step": 1048
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2930592595416846e-05,
+      "loss": 0.332,
+      "step": 1049
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2918090474176378e-05,
+      "loss": 0.3353,
+      "step": 1050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905583365344195e-05,
+      "loss": 0.3321,
+      "step": 1051
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289307129029741e-05,
+      "loss": 0.3279,
+      "step": 1052
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288055427042163e-05,
+      "loss": 0.3364,
+      "step": 1053
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.3251,
+      "step": 1054
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28555054817677e-05,
+      "loss": 0.3249,
+      "step": 1055
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842973755802872e-05,
+      "loss": 0.3263,
+      "step": 1056
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830437170635603e-05,
+      "loss": 0.3346,
+      "step": 1057
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281789574769339e-05,
+      "loss": 0.3298,
+      "step": 1058
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2805349508411996e-05,
+      "loss": 0.3263,
+      "step": 1059
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2792798474235417e-05,
+      "loss": 0.3433,
+      "step": 1060
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780242666615843e-05,
+      "loss": 0.326,
+      "step": 1061
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2767682107013626e-05,
+      "loss": 0.3471,
+      "step": 1062
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755116816897236e-05,
+      "loss": 0.334,
+      "step": 1063
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2742546817743228e-05,
+      "loss": 0.3334,
+      "step": 1064
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2729972131036212e-05,
+      "loss": 0.3263,
+      "step": 1065
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717392778268804e-05,
+      "loss": 0.3477,
+      "step": 1066
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704808780941598e-05,
+      "loss": 0.3292,
+      "step": 1067
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692220160563125e-05,
+      "loss": 0.336,
+      "step": 1068
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2679626938649819e-05,
+      "loss": 0.3516,
+      "step": 1069
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2667029136725975e-05,
+      "loss": 0.3386,
+      "step": 1070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654426776323719e-05,
+      "loss": 0.3133,
+      "step": 1071
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2641819878982973e-05,
+      "loss": 0.3366,
+      "step": 1072
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629208466251406e-05,
+      "loss": 0.3429,
+      "step": 1073
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616592559684408e-05,
+      "loss": 0.3317,
+      "step": 1074
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2603972180845053e-05,
+      "loss": 0.3225,
+      "step": 1075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2591347351304048e-05,
+      "loss": 0.3535,
+      "step": 1076
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2578718092639724e-05,
+      "loss": 0.3349,
+      "step": 1077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2566084426437966e-05,
+      "loss": 0.3305,
+      "step": 1078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553446374292204e-05,
+      "loss": 0.3331,
+      "step": 1079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540803957803356e-05,
+      "loss": 0.3393,
+      "step": 1080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528157198579807e-05,
+      "loss": 0.3257,
+      "step": 1081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251550611823736e-05,
+      "loss": 0.3109,
+      "step": 1082
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.25028507383992e-05,
+      "loss": 0.3239,
+      "step": 1083
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490191080695873e-05,
+      "loss": 0.3629,
+      "step": 1084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477527166765221e-05,
+      "loss": 0.3309,
+      "step": 1085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2464859018252377e-05,
+      "loss": 0.3301,
+      "step": 1086
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2452186656809696e-05,
+      "loss": 0.306,
+      "step": 1087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2439510104096745e-05,
+      "loss": 0.3153,
+      "step": 1088
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242682938178025e-05,
+      "loss": 0.3105,
+      "step": 1089
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414144511534063e-05,
+      "loss": 0.3158,
+      "step": 1090
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2401455515039129e-05,
+      "loss": 0.3137,
+      "step": 1091
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2388762413983447e-05,
+      "loss": 0.3325,
+      "step": 1092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376065230062022e-05,
+      "loss": 0.3416,
+      "step": 1093
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2363363984976848e-05,
+      "loss": 0.3141,
+      "step": 1094
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2350658700436852e-05,
+      "loss": 0.3413,
+      "step": 1095
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2337949398157874e-05,
+      "loss": 0.3294,
+      "step": 1096
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325236099862615e-05,
+      "loss": 0.3136,
+      "step": 1097
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2312518827280603e-05,
+      "loss": 0.3155,
+      "step": 1098
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299797602148167e-05,
+      "loss": 0.2997,
+      "step": 1099
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287072446208388e-05,
+      "loss": 0.3394,
+      "step": 1100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2274343381211067e-05,
+      "loss": 0.3075,
+      "step": 1101
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261610428912679e-05,
+      "loss": 0.3215,
+      "step": 1102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2248873611076351e-05,
+      "loss": 0.3255,
+      "step": 1103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223613294947182e-05,
+      "loss": 0.3207,
+      "step": 1104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2223388465875383e-05,
+      "loss": 0.2916,
+      "step": 1105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2210640182069873e-05,
+      "loss": 0.3232,
+      "step": 1106
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197888119844623e-05,
+      "loss": 0.3123,
+      "step": 1107
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218513230099542e-05,
+      "loss": 0.3545,
+      "step": 1108
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172372747324466e-05,
+      "loss": 0.3195,
+      "step": 1109
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2159609480640361e-05,
+      "loss": 0.3208,
+      "step": 1110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2146842522758042e-05,
+      "loss": 0.3246,
+      "step": 1111
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2134071895498751e-05,
+      "loss": 0.328,
+      "step": 1112
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121297620690011e-05,
+      "loss": 0.3443,
+      "step": 1113
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108519720165575e-05,
+      "loss": 0.3316,
+      "step": 1114
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095738215765391e-05,
+      "loss": 0.3187,
+      "step": 1115
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208295312933557e-05,
+      "loss": 0.3124,
+      "step": 1116
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2070164482728346e-05,
+      "loss": 0.3152,
+      "step": 1117
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2057372297802035e-05,
+      "loss": 0.3041,
+      "step": 1118
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044576596421003e-05,
+      "loss": 0.3143,
+      "step": 1119
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2031777400455625e-05,
+      "loss": 0.3146,
+      "step": 1120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018974731782246e-05,
+      "loss": 0.3245,
+      "step": 1121
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006168612283158e-05,
+      "loss": 0.3311,
+      "step": 1122
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1993359063846535e-05,
+      "loss": 0.3453,
+      "step": 1123
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980546108366425e-05,
+      "loss": 0.3292,
+      "step": 1124
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967729767742688e-05,
+      "loss": 0.3284,
+      "step": 1125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954910063880982e-05,
+      "loss": 0.3353,
+      "step": 1126
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942087018692706e-05,
+      "loss": 0.3324,
+      "step": 1127
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192926065409497e-05,
+      "loss": 0.3098,
+      "step": 1128
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191643099201056e-05,
+      "loss": 0.3349,
+      "step": 1129
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903598054367896e-05,
+      "loss": 0.3236,
+      "step": 1130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890761863100994e-05,
+      "loss": 0.3195,
+      "step": 1131
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877922440149445e-05,
+      "loss": 0.3228,
+      "step": 1132
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865079807458347e-05,
+      "loss": 0.3285,
+      "step": 1133
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852233986978286e-05,
+      "loss": 0.3435,
+      "step": 1134
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1839385000665307e-05,
+      "loss": 0.3405,
+      "step": 1135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1826532870480857e-05,
+      "loss": 0.3011,
+      "step": 1136
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813677618391759e-05,
+      "loss": 0.3135,
+      "step": 1137
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1800819266370175e-05,
+      "loss": 0.3488,
+      "step": 1138
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1787957836393565e-05,
+      "loss": 0.3333,
+      "step": 1139
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775093350444638e-05,
+      "loss": 0.3067,
+      "step": 1140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1762225830511345e-05,
+      "loss": 0.2978,
+      "step": 1141
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749355298586807e-05,
+      "loss": 0.3094,
+      "step": 1142
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.3096,
+      "step": 1143
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1723605286762221e-05,
+      "loss": 0.3174,
+      "step": 1144
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710725850874018e-05,
+      "loss": 0.3223,
+      "step": 1145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697843491018189e-05,
+      "loss": 0.3249,
+      "step": 1146
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1684958229213218e-05,
+      "loss": 0.3114,
+      "step": 1147
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167207008748257e-05,
+      "loss": 0.3287,
+      "step": 1148
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165917908785461e-05,
+      "loss": 0.3182,
+      "step": 1149
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1646285252362599e-05,
+      "loss": 0.3313,
+      "step": 1150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633388603044646e-05,
+      "loss": 0.3078,
+      "step": 1151
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1620489161943665e-05,
+      "loss": 0.3224,
+      "step": 1152
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1607586951107339e-05,
+      "loss": 0.2953,
+      "step": 1153
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1594681992588092e-05,
+      "loss": 0.3133,
+      "step": 1154
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1581774308443042e-05,
+      "loss": 0.3022,
+      "step": 1155
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156886392073396e-05,
+      "loss": 0.3266,
+      "step": 1156
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555950851527256e-05,
+      "loss": 0.3193,
+      "step": 1157
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543035122893898e-05,
+      "loss": 0.3103,
+      "step": 1158
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530116756909415e-05,
+      "loss": 0.3199,
+      "step": 1159
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.151719577565384e-05,
+      "loss": 0.3313,
+      "step": 1160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150427220121168e-05,
+      "loss": 0.3691,
+      "step": 1161
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491346055671866e-05,
+      "loss": 0.3312,
+      "step": 1162
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1478417361127726e-05,
+      "loss": 0.3139,
+      "step": 1163
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465486139676955e-05,
+      "loss": 0.3196,
+      "step": 1164
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452552413421558e-05,
+      "loss": 0.3207,
+      "step": 1165
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1439616204467814e-05,
+      "loss": 0.3312,
+      "step": 1166
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426677534926259e-05,
+      "loss": 0.3065,
+      "step": 1167
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413736426911628e-05,
+      "loss": 0.3146,
+      "step": 1168
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1400792902542827e-05,
+      "loss": 0.3177,
+      "step": 1169
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138784698394289e-05,
+      "loss": 0.3214,
+      "step": 1170
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374898693238943e-05,
+      "loss": 0.3196,
+      "step": 1171
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1361948052562162e-05,
+      "loss": 0.2932,
+      "step": 1172
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134899508404775e-05,
+      "loss": 0.3378,
+      "step": 1173
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336039809834882e-05,
+      "loss": 0.3129,
+      "step": 1174
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132308225206667e-05,
+      "loss": 0.3137,
+      "step": 1175
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131012243289014e-05,
+      "loss": 0.3363,
+      "step": 1176
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1297160374456176e-05,
+      "loss": 0.33,
+      "step": 1177
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128419609891949e-05,
+      "loss": 0.3243,
+      "step": 1178
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271229628438578e-05,
+      "loss": 0.2908,
+      "step": 1179
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1258260985175699e-05,
+      "loss": 0.3281,
+      "step": 1180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245290191296823e-05,
+      "loss": 0.3128,
+      "step": 1181
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1232317268971586e-05,
+      "loss": 0.3281,
+      "step": 1182
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1219342240373276e-05,
+      "loss": 0.3146,
+      "step": 1183
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206365127678771e-05,
+      "loss": 0.3065,
+      "step": 1184
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193385953068512e-05,
+      "loss": 0.2986,
+      "step": 1185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180404738726468e-05,
+      "loss": 0.3098,
+      "step": 1186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167421506840092e-05,
+      "loss": 0.3264,
+      "step": 1187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154436279600287e-05,
+      "loss": 0.3176,
+      "step": 1188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141449079201367e-05,
+      "loss": 0.3134,
+      "step": 1189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128459927841013e-05,
+      "loss": 0.3455,
+      "step": 1190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1115468847720245e-05,
+      "loss": 0.3195,
+      "step": 1191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102475861043385e-05,
+      "loss": 0.3066,
+      "step": 1192
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089480990018e-05,
+      "loss": 0.3096,
+      "step": 1193
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1076484256854889e-05,
+      "loss": 0.3176,
+      "step": 1194
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1063485683768033e-05,
+      "loss": 0.3204,
+      "step": 1195
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050485292974555e-05,
+      "loss": 0.301,
+      "step": 1196
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1037483106694681e-05,
+      "loss": 0.3475,
+      "step": 1197
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102447914715171e-05,
+      "loss": 0.2982,
+      "step": 1198
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1011473436571976e-05,
+      "loss": 0.3185,
+      "step": 1199
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998465997184798e-05,
+      "loss": 0.3277,
+      "step": 1200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098545685122245e-05,
+      "loss": 0.3044,
+      "step": 1201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0972446020920136e-05,
+      "loss": 0.3504,
+      "step": 1202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0959433528515922e-05,
+      "loss": 0.3335,
+      "step": 1203
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094641939625072e-05,
+      "loss": 0.3057,
+      "step": 1204
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0933403646368254e-05,
+      "loss": 0.3097,
+      "step": 1205
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0920386301115e-05,
+      "loss": 0.3407,
+      "step": 1206
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0907367382740168e-05,
+      "loss": 0.3055,
+      "step": 1207
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089434691349566e-05,
+      "loss": 0.3243,
+      "step": 1208
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088132491563602e-05,
+      "loss": 0.3182,
+      "step": 1209
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868301411418405e-05,
+      "loss": 0.3201,
+      "step": 1210
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0855276423102556e-05,
+      "loss": 0.322,
+      "step": 1211
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842249972950743e-05,
+      "loss": 0.3281,
+      "step": 1212
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829222083227738e-05,
+      "loss": 0.3238,
+      "step": 1213
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816192776200767e-05,
+      "loss": 0.3045,
+      "step": 1214
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803162074139489e-05,
+      "loss": 0.2944,
+      "step": 1215
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790129999315942e-05,
+      "loss": 0.3097,
+      "step": 1216
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777096574004502e-05,
+      "loss": 0.3124,
+      "step": 1217
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0764061820481872e-05,
+      "loss": 0.329,
+      "step": 1218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.075102576102701e-05,
+      "loss": 0.3121,
+      "step": 1219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0737988417921108e-05,
+      "loss": 0.3134,
+      "step": 1220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0724949813447563e-05,
+      "loss": 0.3273,
+      "step": 1221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0711909969891915e-05,
+      "loss": 0.2987,
+      "step": 1222
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069886890954182e-05,
+      "loss": 0.3167,
+      "step": 1223
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.068582665468703e-05,
+      "loss": 0.3164,
+      "step": 1224
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0672783227619326e-05,
+      "loss": 0.341,
+      "step": 1225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0659738650632491e-05,
+      "loss": 0.3349,
+      "step": 1226
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646692946022285e-05,
+      "loss": 0.3154,
+      "step": 1227
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633646136086384e-05,
+      "loss": 0.3161,
+      "step": 1228
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0620598243124356e-05,
+      "loss": 0.3126,
+      "step": 1229
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607549289437626e-05,
+      "loss": 0.3384,
+      "step": 1230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594499297329424e-05,
+      "loss": 0.3237,
+      "step": 1231
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.3149,
+      "step": 1232
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0568396287070377e-05,
+      "loss": 0.3045,
+      "step": 1233
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555343313534722e-05,
+      "loss": 0.3182,
+      "step": 1234
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.05422893908079e-05,
+      "loss": 0.3308,
+      "step": 1235
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529234541201631e-05,
+      "loss": 0.2882,
+      "step": 1236
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516178787029234e-05,
+      "loss": 0.301,
+      "step": 1237
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050312215060556e-05,
+      "loss": 0.3177,
+      "step": 1238
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490064654246976e-05,
+      "loss": 0.3061,
+      "step": 1239
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0477006320271317e-05,
+      "loss": 0.331,
+      "step": 1240
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0463947170997848e-05,
+      "loss": 0.3138,
+      "step": 1241
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0450887228747229e-05,
+      "loss": 0.3106,
+      "step": 1242
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437826515841479e-05,
+      "loss": 0.3253,
+      "step": 1243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0424765054603923e-05,
+      "loss": 0.3115,
+      "step": 1244
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041170286735918e-05,
+      "loss": 0.3359,
+      "step": 1245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398639976433094e-05,
+      "loss": 0.3265,
+      "step": 1246
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385576404152728e-05,
+      "loss": 0.3181,
+      "step": 1247
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0372512172846296e-05,
+      "loss": 0.2998,
+      "step": 1248
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035944730484315e-05,
+      "loss": 0.2937,
+      "step": 1249
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034638182247372e-05,
+      "loss": 0.2928,
+      "step": 1250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033331574806949e-05,
+      "loss": 0.3348,
+      "step": 1251
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320249103962954e-05,
+      "loss": 0.3133,
+      "step": 1252
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030718191248759e-05,
+      "loss": 0.3043,
+      "step": 1253
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0294114195977796e-05,
+      "loss": 0.3212,
+      "step": 1254
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0281045976768876e-05,
+      "loss": 0.3182,
+      "step": 1255
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0267977277196995e-05,
+      "loss": 0.3228,
+      "step": 1256
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254908119599134e-05,
+      "loss": 0.2911,
+      "step": 1257
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024183852631306e-05,
+      "loss": 0.3178,
+      "step": 1258
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0228768519677282e-05,
+      "loss": 0.3358,
+      "step": 1259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0215698122031021e-05,
+      "loss": 0.3284,
+      "step": 1260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202627355714156e-05,
+      "loss": 0.3139,
+      "step": 1261
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0189556243067212e-05,
+      "loss": 0.3142,
+      "step": 1262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176484806431288e-05,
+      "loss": 0.3053,
+      "step": 1263
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0163413068148046e-05,
+      "loss": 0.352,
+      "step": 1264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150341050559668e-05,
+      "loss": 0.3205,
+      "step": 1265
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137268776008809e-05,
+      "loss": 0.3131,
+      "step": 1266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0124196266838553e-05,
+      "loss": 0.3072,
+      "step": 1267
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0111123545392402e-05,
+      "loss": 0.2978,
+      "step": 1268
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098050634014216e-05,
+      "loss": 0.3361,
+      "step": 1269
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0084977555048167e-05,
+      "loss": 0.3114,
+      "step": 1270
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071904330838732e-05,
+      "loss": 0.3449,
+      "step": 1271
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058830983730622e-05,
+      "loss": 0.2884,
+      "step": 1272
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045757536068772e-05,
+      "loss": 0.2915,
+      "step": 1273
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003268401019827e-05,
+      "loss": 0.3213,
+      "step": 1274
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019610428464354e-05,
+      "loss": 0.3332,
+      "step": 1275
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006536813212353e-05,
+      "loss": 0.3273,
+      "step": 1276
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99346318678765e-06,
+      "loss": 0.3163,
+      "step": 1277
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980389571535647e-06,
+      "loss": 0.324,
+      "step": 1278
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96731598980173e-06,
+      "loss": 0.2935,
+      "step": 1279
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954242463931235e-06,
+      "loss": 0.3261,
+      "step": 1280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94116901626938e-06,
+      "loss": 0.3053,
+      "step": 1281
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92809566916127e-06,
+      "loss": 0.3085,
+      "step": 1282
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.915022444951836e-06,
+      "loss": 0.3085,
+      "step": 1283
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901949365985787e-06,
+      "loss": 0.3169,
+      "step": 1284
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8888764546076e-06,
+      "loss": 0.3163,
+      "step": 1285
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.875803733161449e-06,
+      "loss": 0.3178,
+      "step": 1286
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862731223991196e-06,
+      "loss": 0.2981,
+      "step": 1287
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84965894944033e-06,
+      "loss": 0.3278,
+      "step": 1288
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.836586931851955e-06,
+      "loss": 0.3068,
+      "step": 1289
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823515193568715e-06,
+      "loss": 0.3136,
+      "step": 1290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81044375693279e-06,
+      "loss": 0.307,
+      "step": 1291
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797372644285846e-06,
+      "loss": 0.3109,
+      "step": 1292
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.784301877968982e-06,
+      "loss": 0.3279,
+      "step": 1293
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.771231480322718e-06,
+      "loss": 0.294,
+      "step": 1294
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758161473686942e-06,
+      "loss": 0.3314,
+      "step": 1295
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74509188040087e-06,
+      "loss": 0.2915,
+      "step": 1296
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73202272280301e-06,
+      "loss": 0.3035,
+      "step": 1297
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.718954023231128e-06,
+      "loss": 0.2886,
+      "step": 1298
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.705885804022207e-06,
+      "loss": 0.3286,
+      "step": 1299
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.692818087512417e-06,
+      "loss": 0.3101,
+      "step": 1300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.679750896037047e-06,
+      "loss": 0.3164,
+      "step": 1301
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666684251930514e-06,
+      "loss": 0.3411,
+      "step": 1302
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.653618177526286e-06,
+      "loss": 0.3192,
+      "step": 1303
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640552695156854e-06,
+      "loss": 0.3082,
+      "step": 1304
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627487827153704e-06,
+      "loss": 0.2961,
+      "step": 1305
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614423595847277e-06,
+      "loss": 0.308,
+      "step": 1306
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60136002356691e-06,
+      "loss": 0.3339,
+      "step": 1307
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588297132640824e-06,
+      "loss": 0.3077,
+      "step": 1308
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57523494539608e-06,
+      "loss": 0.3117,
+      "step": 1309
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.562173484158524e-06,
+      "loss": 0.3055,
+      "step": 1310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549112771252771e-06,
+      "loss": 0.2934,
+      "step": 1311
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536052829002156e-06,
+      "loss": 0.3014,
+      "step": 1312
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.522993679728686e-06,
+      "loss": 0.322,
+      "step": 1313
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.509935345753026e-06,
+      "loss": 0.3041,
+      "step": 1314
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496877849394444e-06,
+      "loss": 0.3162,
+      "step": 1315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48382121297077e-06,
+      "loss": 0.3192,
+      "step": 1316
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470765458798369e-06,
+      "loss": 0.3133,
+      "step": 1317
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.457710609192106e-06,
+      "loss": 0.2902,
+      "step": 1318
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44465668646528e-06,
+      "loss": 0.3161,
+      "step": 1319
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.431603712929623e-06,
+      "loss": 0.332,
+      "step": 1320
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.3078,
+      "step": 1321
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405500702670581e-06,
+      "loss": 0.3206,
+      "step": 1322
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392450710562377e-06,
+      "loss": 0.3126,
+      "step": 1323
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379401756875646e-06,
+      "loss": 0.3105,
+      "step": 1324
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366353863913618e-06,
+      "loss": 0.3093,
+      "step": 1325
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.353307053977717e-06,
+      "loss": 0.3402,
+      "step": 1326
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340261349367512e-06,
+      "loss": 0.3215,
+      "step": 1327
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.327216772380677e-06,
+      "loss": 0.3196,
+      "step": 1328
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314173345312972e-06,
+      "loss": 0.3064,
+      "step": 1329
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301131090458184e-06,
+      "loss": 0.3185,
+      "step": 1330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28809003010809e-06,
+      "loss": 0.3157,
+      "step": 1331
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27505018655244e-06,
+      "loss": 0.3036,
+      "step": 1332
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262011582078894e-06,
+      "loss": 0.327,
+      "step": 1333
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.248974238972993e-06,
+      "loss": 0.3064,
+      "step": 1334
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235938179518131e-06,
+      "loss": 0.3179,
+      "step": 1335
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.2229034259955e-06,
+      "loss": 0.306,
+      "step": 1336
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209870000684063e-06,
+      "loss": 0.3118,
+      "step": 1337
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196837925860516e-06,
+      "loss": 0.3233,
+      "step": 1338
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183807223799236e-06,
+      "loss": 0.2885,
+      "step": 1339
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170777916772266e-06,
+      "loss": 0.285,
+      "step": 1340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15775002704926e-06,
+      "loss": 0.3004,
+      "step": 1341
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.144723576897447e-06,
+      "loss": 0.3024,
+      "step": 1342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.131698588581596e-06,
+      "loss": 0.2974,
+      "step": 1343
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118675084363986e-06,
+      "loss": 0.3008,
+      "step": 1344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105653086504342e-06,
+      "loss": 0.357,
+      "step": 1345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092632617259832e-06,
+      "loss": 0.3076,
+      "step": 1346
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079613698885002e-06,
+      "loss": 0.2938,
+      "step": 1347
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.066596353631749e-06,
+      "loss": 0.3094,
+      "step": 1348
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.053580603749282e-06,
+      "loss": 0.3126,
+      "step": 1349
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.040566471484085e-06,
+      "loss": 0.2992,
+      "step": 1350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.027553979079869e-06,
+      "loss": 0.3113,
+      "step": 1351
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01454314877755e-06,
+      "loss": 0.29,
+      "step": 1352
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001534002815209e-06,
+      "loss": 0.3537,
+      "step": 1353
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.988526563428029e-06,
+      "loss": 0.2985,
+      "step": 1354
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.975520852848291e-06,
+      "loss": 0.3452,
+      "step": 1355
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962516893305324e-06,
+      "loss": 0.2938,
+      "step": 1356
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.949514707025448e-06,
+      "loss": 0.3278,
+      "step": 1357
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.936514316231965e-06,
+      "loss": 0.2946,
+      "step": 1358
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.923515743145113e-06,
+      "loss": 0.2937,
+      "step": 1359
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910519009982003e-06,
+      "loss": 0.2854,
+      "step": 1360
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897524138956619e-06,
+      "loss": 0.3222,
+      "step": 1361
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.884531152279757e-06,
+      "loss": 0.3112,
+      "step": 1362
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87154007215899e-06,
+      "loss": 0.3293,
+      "step": 1363
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858550920798634e-06,
+      "loss": 0.2971,
+      "step": 1364
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845563720399715e-06,
+      "loss": 0.3078,
+      "step": 1365
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83257849315991e-06,
+      "loss": 0.2865,
+      "step": 1366
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819595261273532e-06,
+      "loss": 0.3144,
+      "step": 1367
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806614046931491e-06,
+      "loss": 0.3299,
+      "step": 1368
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.793634872321232e-06,
+      "loss": 0.3067,
+      "step": 1369
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780657759626728e-06,
+      "loss": 0.3308,
+      "step": 1370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.767682731028415e-06,
+      "loss": 0.3435,
+      "step": 1371
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754709808703179e-06,
+      "loss": 0.2929,
+      "step": 1372
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.741739014824303e-06,
+      "loss": 0.3138,
+      "step": 1373
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728770371561424e-06,
+      "loss": 0.3094,
+      "step": 1374
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.715803901080514e-06,
+      "loss": 0.3134,
+      "step": 1375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.702839625543829e-06,
+      "loss": 0.338,
+      "step": 1376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.689877567109861e-06,
+      "loss": 0.3203,
+      "step": 1377
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67691774793333e-06,
+      "loss": 0.3254,
+      "step": 1378
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.663960190165123e-06,
+      "loss": 0.3155,
+      "step": 1379
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.651004915952252e-06,
+      "loss": 0.3051,
+      "step": 1380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63805194743784e-06,
+      "loss": 0.3089,
+      "step": 1381
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625101306761063e-06,
+      "loss": 0.3241,
+      "step": 1382
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.612153016057114e-06,
+      "loss": 0.3115,
+      "step": 1383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599207097457175e-06,
+      "loss": 0.3072,
+      "step": 1384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.586263573088374e-06,
+      "loss": 0.3114,
+      "step": 1385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573322465073746e-06,
+      "loss": 0.3162,
+      "step": 1386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56038379553219e-06,
+      "loss": 0.3025,
+      "step": 1387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547447586578448e-06,
+      "loss": 0.2864,
+      "step": 1388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534513860323047e-06,
+      "loss": 0.3234,
+      "step": 1389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521582638872273e-06,
+      "loss": 0.3024,
+      "step": 1390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50865394432814e-06,
+      "loss": 0.3039,
+      "step": 1391
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.495727798788323e-06,
+      "loss": 0.2841,
+      "step": 1392
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48280422434616e-06,
+      "loss": 0.2938,
+      "step": 1393
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469883243090589e-06,
+      "loss": 0.3207,
+      "step": 1394
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456964877106104e-06,
+      "loss": 0.3075,
+      "step": 1395
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444049148472744e-06,
+      "loss": 0.2826,
+      "step": 1396
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431136079266041e-06,
+      "loss": 0.2984,
+      "step": 1397
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418225691556962e-06,
+      "loss": 0.3225,
+      "step": 1398
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.405318007411911e-06,
+      "loss": 0.3014,
+      "step": 1399
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.392413048892665e-06,
+      "loss": 0.2908,
+      "step": 1400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.379510838056338e-06,
+      "loss": 0.3088,
+      "step": 1401
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366611396955352e-06,
+      "loss": 0.2823,
+      "step": 1402
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.353714747637403e-06,
+      "loss": 0.296,
+      "step": 1403
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.340820912145391e-06,
+      "loss": 0.3205,
+      "step": 1404
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.327929912517435e-06,
+      "loss": 0.3178,
+      "step": 1405
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315041770786783e-06,
+      "loss": 0.282,
+      "step": 1406
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302156508981816e-06,
+      "loss": 0.2936,
+      "step": 1407
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289274149125987e-06,
+      "loss": 0.3004,
+      "step": 1408
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276394713237782e-06,
+      "loss": 0.2898,
+      "step": 1409
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.3125,
+      "step": 1410
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.250644701413197e-06,
+      "loss": 0.2723,
+      "step": 1411
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237774169488658e-06,
+      "loss": 0.3171,
+      "step": 1412
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224906649555365e-06,
+      "loss": 0.2989,
+      "step": 1413
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21204216360644e-06,
+      "loss": 0.2686,
+      "step": 1414
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.199180733629826e-06,
+      "loss": 0.2877,
+      "step": 1415
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18632238160824e-06,
+      "loss": 0.3083,
+      "step": 1416
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.173467129519148e-06,
+      "loss": 0.302,
+      "step": 1417
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.160614999334697e-06,
+      "loss": 0.309,
+      "step": 1418
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147766013021716e-06,
+      "loss": 0.2942,
+      "step": 1419
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13492019254166e-06,
+      "loss": 0.3097,
+      "step": 1420
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122077559850557e-06,
+      "loss": 0.2772,
+      "step": 1421
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109238136899004e-06,
+      "loss": 0.2912,
+      "step": 1422
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09640194563211e-06,
+      "loss": 0.2907,
+      "step": 1423
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083569007989445e-06,
+      "loss": 0.3096,
+      "step": 1424
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.070739345905032e-06,
+      "loss": 0.3116,
+      "step": 1425
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.057912981307298e-06,
+      "loss": 0.3115,
+      "step": 1426
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04508993611902e-06,
+      "loss": 0.2758,
+      "step": 1427
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032270232257312e-06,
+      "loss": 0.3115,
+      "step": 1428
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01945389163358e-06,
+      "loss": 0.3222,
+      "step": 1429
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006640936153466e-06,
+      "loss": 0.298,
+      "step": 1430
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993831387716844e-06,
+      "loss": 0.3097,
+      "step": 1431
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981025268217755e-06,
+      "loss": 0.2821,
+      "step": 1432
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.968222599544377e-06,
+      "loss": 0.3093,
+      "step": 1433
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955423403578998e-06,
+      "loss": 0.3304,
+      "step": 1434
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942627702197968e-06,
+      "loss": 0.299,
+      "step": 1435
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.929835517271656e-06,
+      "loss": 0.3236,
+      "step": 1436
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917046870664431e-06,
+      "loss": 0.27,
+      "step": 1437
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.904261784234612e-06,
+      "loss": 0.3076,
+      "step": 1438
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.891480279834426e-06,
+      "loss": 0.3034,
+      "step": 1439
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878702379309992e-06,
+      "loss": 0.3232,
+      "step": 1440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.865928104501252e-06,
+      "loss": 0.299,
+      "step": 1441
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.853157477241961e-06,
+      "loss": 0.2905,
+      "step": 1442
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.840390519359644e-06,
+      "loss": 0.2928,
+      "step": 1443
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827627252675537e-06,
+      "loss": 0.3195,
+      "step": 1444
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814867699004585e-06,
+      "loss": 0.2981,
+      "step": 1445
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802111880155382e-06,
+      "loss": 0.274,
+      "step": 1446
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.789359817930132e-06,
+      "loss": 0.3072,
+      "step": 1447
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77661153412462e-06,
+      "loss": 0.2753,
+      "step": 1448
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.763867050528184e-06,
+      "loss": 0.3104,
+      "step": 1449
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.75112638892365e-06,
+      "loss": 0.3115,
+      "step": 1450
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738389571087325e-06,
+      "loss": 0.3219,
+      "step": 1451
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.725656618788938e-06,
+      "loss": 0.3326,
+      "step": 1452
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712927553791614e-06,
+      "loss": 0.2955,
+      "step": 1453
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700202397851834e-06,
+      "loss": 0.2693,
+      "step": 1454
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.687481172719402e-06,
+      "loss": 0.3046,
+      "step": 1455
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67476390013739e-06,
+      "loss": 0.2953,
+      "step": 1456
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662050601842128e-06,
+      "loss": 0.292,
+      "step": 1457
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.649341299563151e-06,
+      "loss": 0.3194,
+      "step": 1458
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.636636015023155e-06,
+      "loss": 0.3008,
+      "step": 1459
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62393476993798e-06,
+      "loss": 0.3137,
+      "step": 1460
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.611237586016558e-06,
+      "loss": 0.3102,
+      "step": 1461
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.598544484960873e-06,
+      "loss": 0.3041,
+      "step": 1462
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585855488465938e-06,
+      "loss": 0.2915,
+      "step": 1463
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.573170618219754e-06,
+      "loss": 0.3169,
+      "step": 1464
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.560489895903258e-06,
+      "loss": 0.2969,
+      "step": 1465
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.547813343190307e-06,
+      "loss": 0.3159,
+      "step": 1466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535140981747627e-06,
+      "loss": 0.2994,
+      "step": 1467
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.522472833234779e-06,
+      "loss": 0.3229,
+      "step": 1468
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.509808919304129e-06,
+      "loss": 0.317,
+      "step": 1469
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497149261600803e-06,
+      "loss": 0.3,
+      "step": 1470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484493881762643e-06,
+      "loss": 0.3079,
+      "step": 1471
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4718428014201935e-06,
+      "loss": 0.2801,
+      "step": 1472
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.459196042196647e-06,
+      "loss": 0.2965,
+      "step": 1473
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4465536257078e-06,
+      "loss": 0.3139,
+      "step": 1474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433915573562035e-06,
+      "loss": 0.2807,
+      "step": 1475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42128190736028e-06,
+      "loss": 0.2975,
+      "step": 1476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4086526486959545e-06,
+      "loss": 0.3204,
+      "step": 1477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.396027819154954e-06,
+      "loss": 0.3065,
+      "step": 1478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383407440315595e-06,
+      "loss": 0.292,
+      "step": 1479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370791533748596e-06,
+      "loss": 0.2971,
+      "step": 1480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3581801210170325e-06,
+      "loss": 0.3164,
+      "step": 1481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345573223676284e-06,
+      "loss": 0.3075,
+      "step": 1482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.332970863274029e-06,
+      "loss": 0.335,
+      "step": 1483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320373061350188e-06,
+      "loss": 0.3179,
+      "step": 1484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307779839436878e-06,
+      "loss": 0.3139,
+      "step": 1485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2951912190584026e-06,
+      "loss": 0.2842,
+      "step": 1486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2826072217312e-06,
+      "loss": 0.316,
+      "step": 1487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.27002786896379e-06,
+      "loss": 0.2873,
+      "step": 1488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.257453182256773e-06,
+      "loss": 0.3223,
+      "step": 1489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244883183102769e-06,
+      "loss": 0.3195,
+      "step": 1490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.232317892986376e-06,
+      "loss": 0.2981,
+      "step": 1491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219757333384157e-06,
+      "loss": 0.3343,
+      "step": 1492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.207201525764587e-06,
+      "loss": 0.3359,
+      "step": 1493
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.194650491588007e-06,
+      "loss": 0.3108,
+      "step": 1494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182104252306612e-06,
+      "loss": 0.292,
+      "step": 1495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169562829364402e-06,
+      "loss": 0.2871,
+      "step": 1496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157026244197132e-06,
+      "loss": 0.2941,
+      "step": 1497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1444945182322985e-06,
+      "loss": 0.3167,
+      "step": 1498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.3187,
+      "step": 1499
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.119445729578374e-06,
+      "loss": 0.3154,
+      "step": 1500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.10692870970259e-06,
+      "loss": 0.3368,
+      "step": 1501
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.094416634655808e-06,
+      "loss": 0.2989,
+      "step": 1502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.081909525823625e-06,
+      "loss": 0.291,
+      "step": 1503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069407404583157e-06,
+      "loss": 0.2914,
+      "step": 1504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0569102923029945e-06,
+      "loss": 0.2959,
+      "step": 1505
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.044418210343161e-06,
+      "loss": 0.2927,
+      "step": 1506
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031931180055087e-06,
+      "loss": 0.3182,
+      "step": 1507
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.019449222781576e-06,
+      "loss": 0.308,
+      "step": 1508
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006972359856743e-06,
+      "loss": 0.3068,
+      "step": 1509
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994500612606009e-06,
+      "loss": 0.3149,
+      "step": 1510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.982034002346053e-06,
+      "loss": 0.3361,
+      "step": 1511
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.96957255038476e-06,
+      "loss": 0.3044,
+      "step": 1512
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.957116278021212e-06,
+      "loss": 0.2904,
+      "step": 1513
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944665206545637e-06,
+      "loss": 0.3133,
+      "step": 1514
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.932219357239362e-06,
+      "loss": 0.2667,
+      "step": 1515
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9197787513748035e-06,
+      "loss": 0.3302,
+      "step": 1516
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9073434102154035e-06,
+      "loss": 0.3132,
+      "step": 1517
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894913355015611e-06,
+      "loss": 0.3317,
+      "step": 1518
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882488607020848e-06,
+      "loss": 0.2944,
+      "step": 1519
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870069187467448e-06,
+      "loss": 0.3347,
+      "step": 1520
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.857655117582647e-06,
+      "loss": 0.2946,
+      "step": 1521
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.845246418584546e-06,
+      "loss": 0.2884,
+      "step": 1522
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832843111682046e-06,
+      "loss": 0.2903,
+      "step": 1523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.820445218074849e-06,
+      "loss": 0.307,
+      "step": 1524
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808052758953402e-06,
+      "loss": 0.2946,
+      "step": 1525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.79566575549885e-06,
+      "loss": 0.3221,
+      "step": 1526
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783284228883029e-06,
+      "loss": 0.3201,
+      "step": 1527
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770908200268411e-06,
+      "loss": 0.3082,
+      "step": 1528
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7585376908080625e-06,
+      "loss": 0.3031,
+      "step": 1529
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746172721645625e-06,
+      "loss": 0.2903,
+      "step": 1530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7338133139152675e-06,
+      "loss": 0.2876,
+      "step": 1531
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7214594887416506e-06,
+      "loss": 0.295,
+      "step": 1532
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7091112672399e-06,
+      "loss": 0.2835,
+      "step": 1533
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.696768670515561e-06,
+      "loss": 0.3058,
+      "step": 1534
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6844317196645595e-06,
+      "loss": 0.3026,
+      "step": 1535
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.672100435773176e-06,
+      "loss": 0.291,
+      "step": 1536
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.659774839918012e-06,
+      "loss": 0.2734,
+      "step": 1537
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6474549531659285e-06,
+      "loss": 0.3185,
+      "step": 1538
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6351407965740465e-06,
+      "loss": 0.3006,
+      "step": 1539
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622832391189689e-06,
+      "loss": 0.3107,
+      "step": 1540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.610529758050344e-06,
+      "loss": 0.2877,
+      "step": 1541
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5982329181836325e-06,
+      "loss": 0.2948,
+      "step": 1542
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585941892607283e-06,
+      "loss": 0.2712,
+      "step": 1543
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.573656702329075e-06,
+      "loss": 0.3112,
+      "step": 1544
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561377368346824e-06,
+      "loss": 0.3245,
+      "step": 1545
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549103911648333e-06,
+      "loss": 0.3061,
+      "step": 1546
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536836353211353e-06,
+      "loss": 0.302,
+      "step": 1547
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524574714003562e-06,
+      "loss": 0.2841,
+      "step": 1548
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.512319014982522e-06,
+      "loss": 0.2874,
+      "step": 1549
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50006927709563e-06,
+      "loss": 0.2896,
+      "step": 1550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487825521280109e-06,
+      "loss": 0.2957,
+      "step": 1551
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.475587768462945e-06,
+      "loss": 0.3117,
+      "step": 1552
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.463356039560871e-06,
+      "loss": 0.3083,
+      "step": 1553
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451130355480326e-06,
+      "loss": 0.288,
+      "step": 1554
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438910737117409e-06,
+      "loss": 0.2936,
+      "step": 1555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426697205357857e-06,
+      "loss": 0.318,
+      "step": 1556
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.414489781077009e-06,
+      "loss": 0.2835,
+      "step": 1557
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.402288485139748e-06,
+      "loss": 0.2996,
+      "step": 1558
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3900933384005e-06,
+      "loss": 0.2995,
+      "step": 1559
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3779043617031775e-06,
+      "loss": 0.2878,
+      "step": 1560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365721575881135e-06,
+      "loss": 0.272,
+      "step": 1561
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35354500175716e-06,
+      "loss": 0.2835,
+      "step": 1562
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.341374660143419e-06,
+      "loss": 0.3147,
+      "step": 1563
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.329210571841419e-06,
+      "loss": 0.3306,
+      "step": 1564
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317052757641986e-06,
+      "loss": 0.3088,
+      "step": 1565
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.304901238325224e-06,
+      "loss": 0.2893,
+      "step": 1566
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.292756034660472e-06,
+      "loss": 0.2898,
+      "step": 1567
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280617167406273e-06,
+      "loss": 0.2956,
+      "step": 1568
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268484657310351e-06,
+      "loss": 0.3172,
+      "step": 1569
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2563585251095495e-06,
+      "loss": 0.3204,
+      "step": 1570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.244238791529822e-06,
+      "loss": 0.3,
+      "step": 1571
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.232125477286184e-06,
+      "loss": 0.3235,
+      "step": 1572
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.220018603082672e-06,
+      "loss": 0.2853,
+      "step": 1573
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.207918189612321e-06,
+      "loss": 0.2897,
+      "step": 1574
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.195824257557126e-06,
+      "loss": 0.3069,
+      "step": 1575
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183736827587995e-06,
+      "loss": 0.2989,
+      "step": 1576
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.171655920364731e-06,
+      "loss": 0.3135,
+      "step": 1577
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159581556535989e-06,
+      "loss": 0.3225,
+      "step": 1578
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147513756739227e-06,
+      "loss": 0.3004,
+      "step": 1579
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1354525416007e-06,
+      "loss": 0.3062,
+      "step": 1580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123397931735402e-06,
+      "loss": 0.3149,
+      "step": 1581
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.111349947747034e-06,
+      "loss": 0.2845,
+      "step": 1582
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0993086102279746e-06,
+      "loss": 0.2973,
+      "step": 1583
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.08727393975925e-06,
+      "loss": 0.2847,
+      "step": 1584
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0752459569104725e-06,
+      "loss": 0.2975,
+      "step": 1585
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063224682239845e-06,
+      "loss": 0.2822,
+      "step": 1586
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051210136294089e-06,
+      "loss": 0.2861,
+      "step": 1587
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.3196,
+      "step": 1588
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027201312706571e-06,
+      "loss": 0.3166,
+      "step": 1589
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.015207076100618e-06,
+      "loss": 0.2858,
+      "step": 1590
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.003219650291089e-06,
+      "loss": 0.2911,
+      "step": 1591
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.991239055766862e-06,
+      "loss": 0.3049,
+      "step": 1592
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979265313005128e-06,
+      "loss": 0.2895,
+      "step": 1593
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.967298442471379e-06,
+      "loss": 0.3026,
+      "step": 1594
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9553384646193515e-06,
+      "loss": 0.3008,
+      "step": 1595
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.943385399891004e-06,
+      "loss": 0.3173,
+      "step": 1596
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931439268716481e-06,
+      "loss": 0.2932,
+      "step": 1597
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919500091514081e-06,
+      "loss": 0.2756,
+      "step": 1598
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9075678886902e-06,
+      "loss": 0.3011,
+      "step": 1599
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.895642680639328e-06,
+      "loss": 0.2972,
+      "step": 1600
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883724487744004e-06,
+      "loss": 0.2813,
+      "step": 1601
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.871813330374756e-06,
+      "loss": 0.2997,
+      "step": 1602
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.859909228890104e-06,
+      "loss": 0.2945,
+      "step": 1603
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.848012203636511e-06,
+      "loss": 0.3008,
+      "step": 1604
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8361222749483246e-06,
+      "loss": 0.2871,
+      "step": 1605
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824239463147784e-06,
+      "loss": 0.3225,
+      "step": 1606
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.812363788544959e-06,
+      "loss": 0.3018,
+      "step": 1607
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.800495271437712e-06,
+      "loss": 0.3024,
+      "step": 1608
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788633932111678e-06,
+      "loss": 0.2778,
+      "step": 1609
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7767797908402305e-06,
+      "loss": 0.3053,
+      "step": 1610
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.764932867884423e-06,
+      "loss": 0.2856,
+      "step": 1611
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753093183492988e-06,
+      "loss": 0.2741,
+      "step": 1612
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7412607579022775e-06,
+      "loss": 0.3439,
+      "step": 1613
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.729435611336239e-06,
+      "loss": 0.3011,
+      "step": 1614
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7176177640063805e-06,
+      "loss": 0.2898,
+      "step": 1615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.705807236111733e-06,
+      "loss": 0.2741,
+      "step": 1616
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694004047838812e-06,
+      "loss": 0.2853,
+      "step": 1617
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682208219361592e-06,
+      "loss": 0.2897,
+      "step": 1618
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670419770841477e-06,
+      "loss": 0.2584,
+      "step": 1619
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.658638722427237e-06,
+      "loss": 0.2708,
+      "step": 1620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.646865094255012e-06,
+      "loss": 0.2992,
+      "step": 1621
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6350989064482555e-06,
+      "loss": 0.2903,
+      "step": 1622
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6233401791176946e-06,
+      "loss": 0.2808,
+      "step": 1623
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611588932361318e-06,
+      "loss": 0.295,
+      "step": 1624
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599845186264317e-06,
+      "loss": 0.2915,
+      "step": 1625
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588108960899069e-06,
+      "loss": 0.331,
+      "step": 1626
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576380276325102e-06,
+      "loss": 0.2703,
+      "step": 1627
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564659152589041e-06,
+      "loss": 0.2818,
+      "step": 1628
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552945609724601e-06,
+      "loss": 0.3119,
+      "step": 1629
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541239667752538e-06,
+      "loss": 0.2945,
+      "step": 1630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.529541346680607e-06,
+      "loss": 0.2826,
+      "step": 1631
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517850666503547e-06,
+      "loss": 0.3092,
+      "step": 1632
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.50616764720304e-06,
+      "loss": 0.3014,
+      "step": 1633
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494492308747658e-06,
+      "loss": 0.2972,
+      "step": 1634
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.482824671092862e-06,
+      "loss": 0.2936,
+      "step": 1635
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4711647541809475e-06,
+      "loss": 0.31,
+      "step": 1636
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.459512577941002e-06,
+      "loss": 0.2918,
+      "step": 1637
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.447868162288895e-06,
+      "loss": 0.2858,
+      "step": 1638
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436231527127228e-06,
+      "loss": 0.2944,
+      "step": 1639
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4246026923453046e-06,
+      "loss": 0.2815,
+      "step": 1640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412981677819094e-06,
+      "loss": 0.281,
+      "step": 1641
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401368503411203e-06,
+      "loss": 0.2922,
+      "step": 1642
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3897631889708305e-06,
+      "loss": 0.31,
+      "step": 1643
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3781657543337484e-06,
+      "loss": 0.2829,
+      "step": 1644
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366576219322261e-06,
+      "loss": 0.2778,
+      "step": 1645
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354994603745162e-06,
+      "loss": 0.3041,
+      "step": 1646
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.343420927397718e-06,
+      "loss": 0.3022,
+      "step": 1647
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.331855210061627e-06,
+      "loss": 0.288,
+      "step": 1648
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.320297471504972e-06,
+      "loss": 0.3345,
+      "step": 1649
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308747731482207e-06,
+      "loss": 0.2837,
+      "step": 1650
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.297206009734119e-06,
+      "loss": 0.2802,
+      "step": 1651
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2856723259877805e-06,
+      "loss": 0.2929,
+      "step": 1652
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274146699956531e-06,
+      "loss": 0.2412,
+      "step": 1653
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2626291513399395e-06,
+      "loss": 0.2422,
+      "step": 1654
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2511196998237616e-06,
+      "loss": 0.2202,
+      "step": 1655
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.239618365079921e-06,
+      "loss": 0.2186,
+      "step": 1656
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.228125166766469e-06,
+      "loss": 0.2398,
+      "step": 1657
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.21664012452754e-06,
+      "loss": 0.197,
+      "step": 1658
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.205163257993341e-06,
+      "loss": 0.2043,
+      "step": 1659
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1936945867800915e-06,
+      "loss": 0.2273,
+      "step": 1660
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.182234130490014e-06,
+      "loss": 0.2086,
+      "step": 1661
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170781908711289e-06,
+      "loss": 0.2025,
+      "step": 1662
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.159337941018021e-06,
+      "loss": 0.2202,
+      "step": 1663
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.147902246970202e-06,
+      "loss": 0.1927,
+      "step": 1664
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136474846113688e-06,
+      "loss": 0.2051,
+      "step": 1665
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.125055757980162e-06,
+      "loss": 0.2301,
+      "step": 1666
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113645002087093e-06,
+      "loss": 0.2283,
+      "step": 1667
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1022425979377174e-06,
+      "loss": 0.2005,
+      "step": 1668
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.090848565020983e-06,
+      "loss": 0.2183,
+      "step": 1669
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079462922811541e-06,
+      "loss": 0.2148,
+      "step": 1670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068085690769702e-06,
+      "loss": 0.1863,
+      "step": 1671
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.056716888341391e-06,
+      "loss": 0.1921,
+      "step": 1672
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0453565349581324e-06,
+      "loss": 0.1889,
+      "step": 1673
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.034004650037016e-06,
+      "loss": 0.1946,
+      "step": 1674
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022661252980643e-06,
+      "loss": 0.1907,
+      "step": 1675
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011326363177117e-06,
+      "loss": 0.2345,
+      "step": 1676
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.207,
+      "step": 1677
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.988682182808279e-06,
+      "loss": 0.1974,
+      "step": 1678
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97737293094633e-06,
+      "loss": 0.2278,
+      "step": 1679
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.966072263743899e-06,
+      "loss": 0.2197,
+      "step": 1680
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.954780200516047e-06,
+      "loss": 0.2002,
+      "step": 1681
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943496760563139e-06,
+      "loss": 0.1871,
+      "step": 1682
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932221963170801e-06,
+      "loss": 0.1931,
+      "step": 1683
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920955827609876e-06,
+      "loss": 0.2083,
+      "step": 1684
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.909698373136416e-06,
+      "loss": 0.2001,
+      "step": 1685
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89844961899163e-06,
+      "loss": 0.2173,
+      "step": 1686
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.88720958440185e-06,
+      "loss": 0.1976,
+      "step": 1687
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875978288578512e-06,
+      "loss": 0.206,
+      "step": 1688
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8647557507181164e-06,
+      "loss": 0.1913,
+      "step": 1689
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853541990002195e-06,
+      "loss": 0.1999,
+      "step": 1690
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842337025597268e-06,
+      "loss": 0.2105,
+      "step": 1691
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.831140876654831e-06,
+      "loss": 0.198,
+      "step": 1692
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.819953562311308e-06,
+      "loss": 0.1886,
+      "step": 1693
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.808775101688026e-06,
+      "loss": 0.2011,
+      "step": 1694
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797605513891179e-06,
+      "loss": 0.1977,
+      "step": 1695
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7864448180117875e-06,
+      "loss": 0.2064,
+      "step": 1696
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.775293033125687e-06,
+      "loss": 0.2082,
+      "step": 1697
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.764150178293471e-06,
+      "loss": 0.2074,
+      "step": 1698
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753016272560475e-06,
+      "loss": 0.2109,
+      "step": 1699
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.741891334956742e-06,
+      "loss": 0.2398,
+      "step": 1700
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.730775384496976e-06,
+      "loss": 0.1916,
+      "step": 1701
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.719668440180533e-06,
+      "loss": 0.2152,
+      "step": 1702
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.70857052099137e-06,
+      "loss": 0.2053,
+      "step": 1703
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697481645898012e-06,
+      "loss": 0.2018,
+      "step": 1704
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.686401833853536e-06,
+      "loss": 0.2071,
+      "step": 1705
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675331103795528e-06,
+      "loss": 0.1928,
+      "step": 1706
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.66426947464604e-06,
+      "loss": 0.1921,
+      "step": 1707
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653216965311581e-06,
+      "loss": 0.1992,
+      "step": 1708
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642173594683069e-06,
+      "loss": 0.1868,
+      "step": 1709
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.631139381635795e-06,
+      "loss": 0.205,
+      "step": 1710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620114345029406e-06,
+      "loss": 0.1957,
+      "step": 1711
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.609098503707869e-06,
+      "loss": 0.203,
+      "step": 1712
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598091876499417e-06,
+      "loss": 0.2184,
+      "step": 1713
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5870944822165505e-06,
+      "loss": 0.2047,
+      "step": 1714
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.576106339655985e-06,
+      "loss": 0.2007,
+      "step": 1715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565127467598619e-06,
+      "loss": 0.2126,
+      "step": 1716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554157884809513e-06,
+      "loss": 0.2142,
+      "step": 1717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.543197610037839e-06,
+      "loss": 0.2206,
+      "step": 1718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.532246662016872e-06,
+      "loss": 0.196,
+      "step": 1719
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.521305059463936e-06,
+      "loss": 0.2022,
+      "step": 1720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.510372821080394e-06,
+      "loss": 0.2126,
+      "step": 1721
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4994499655515865e-06,
+      "loss": 0.2024,
+      "step": 1722
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48853651154683e-06,
+      "loss": 0.2001,
+      "step": 1723
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.477632477719373e-06,
+      "loss": 0.2043,
+      "step": 1724
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46673788270635e-06,
+      "loss": 0.2064,
+      "step": 1725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455852745128774e-06,
+      "loss": 0.1894,
+      "step": 1726
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4449770835914964e-06,
+      "loss": 0.1995,
+      "step": 1727
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4341109166831565e-06,
+      "loss": 0.195,
+      "step": 1728
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423254262976179e-06,
+      "loss": 0.2063,
+      "step": 1729
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.412407141026729e-06,
+      "loss": 0.2168,
+      "step": 1730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4015695693746685e-06,
+      "loss": 0.1793,
+      "step": 1731
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.390741566543549e-06,
+      "loss": 0.1978,
+      "step": 1732
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.379923151040557e-06,
+      "loss": 0.2284,
+      "step": 1733
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.369114341356497e-06,
+      "loss": 0.2065,
+      "step": 1734
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358315155965762e-06,
+      "loss": 0.2154,
+      "step": 1735
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.347525613326278e-06,
+      "loss": 0.216,
+      "step": 1736
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3367457318795034e-06,
+      "loss": 0.21,
+      "step": 1737
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.325975530050387e-06,
+      "loss": 0.1804,
+      "step": 1738
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.315215026247315e-06,
+      "loss": 0.2028,
+      "step": 1739
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.304464238862115e-06,
+      "loss": 0.2018,
+      "step": 1740
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29372318627e-06,
+      "loss": 0.2331,
+      "step": 1741
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282991886829548e-06,
+      "loss": 0.175,
+      "step": 1742
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.272270358882667e-06,
+      "loss": 0.1809,
+      "step": 1743
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2615586207545524e-06,
+      "loss": 0.2149,
+      "step": 1744
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250856690753682e-06,
+      "loss": 0.1982,
+      "step": 1745
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.240164587171761e-06,
+      "loss": 0.1965,
+      "step": 1746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.229482328283707e-06,
+      "loss": 0.1967,
+      "step": 1747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.218809932347597e-06,
+      "loss": 0.1982,
+      "step": 1748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.208147417604665e-06,
+      "loss": 0.2109,
+      "step": 1749
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.19749480227925e-06,
+      "loss": 0.1986,
+      "step": 1750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186852104578768e-06,
+      "loss": 0.1957,
+      "step": 1751
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.176219342693687e-06,
+      "loss": 0.2208,
+      "step": 1752
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165596534797497e-06,
+      "loss": 0.2008,
+      "step": 1753
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154983699046664e-06,
+      "loss": 0.2117,
+      "step": 1754
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1443808535806195e-06,
+      "loss": 0.2061,
+      "step": 1755
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.133788016521718e-06,
+      "loss": 0.2229,
+      "step": 1756
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.123205205975199e-06,
+      "loss": 0.2053,
+      "step": 1757
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112632440029176e-06,
+      "loss": 0.1969,
+      "step": 1758
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.102069736754593e-06,
+      "loss": 0.1964,
+      "step": 1759
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091517114205185e-06,
+      "loss": 0.1923,
+      "step": 1760
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.080974590417464e-06,
+      "loss": 0.1941,
+      "step": 1761
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0704421834106876e-06,
+      "loss": 0.2154,
+      "step": 1762
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059919911186808e-06,
+      "loss": 0.2018,
+      "step": 1763
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.049407791730464e-06,
+      "loss": 0.2233,
+      "step": 1764
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0389058430089435e-06,
+      "loss": 0.2071,
+      "step": 1765
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.212,
+      "step": 1766
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017932529552543e-06,
+      "loss": 0.196,
+      "step": 1767
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007461200665193e-06,
+      "loss": 0.1903,
+      "step": 1768
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997000114207652e-06,
+      "loss": 0.2022,
+      "step": 1769
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.986549288059985e-06,
+      "loss": 0.2181,
+      "step": 1770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9761087400847066e-06,
+      "loss": 0.2079,
+      "step": 1771
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965678488126773e-06,
+      "loss": 0.2178,
+      "step": 1772
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.955258550013544e-06,
+      "loss": 0.2194,
+      "step": 1773
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944848943554741e-06,
+      "loss": 0.1975,
+      "step": 1774
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.934449686542436e-06,
+      "loss": 0.2097,
+      "step": 1775
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924060796751012e-06,
+      "loss": 0.2088,
+      "step": 1776
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913682291937123e-06,
+      "loss": 0.2103,
+      "step": 1777
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903314189839682e-06,
+      "loss": 0.2025,
+      "step": 1778
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892956508179822e-06,
+      "loss": 0.2021,
+      "step": 1779
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8826092646608556e-06,
+      "loss": 0.2013,
+      "step": 1780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.872272476968265e-06,
+      "loss": 0.2021,
+      "step": 1781
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8619461627696605e-06,
+      "loss": 0.2221,
+      "step": 1782
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.851630339714741e-06,
+      "loss": 0.2085,
+      "step": 1783
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.841325025435286e-06,
+      "loss": 0.2214,
+      "step": 1784
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.83103023754511e-06,
+      "loss": 0.1974,
+      "step": 1785
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820745993640029e-06,
+      "loss": 0.2174,
+      "step": 1786
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.810472311297846e-06,
+      "loss": 0.1986,
+      "step": 1787
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8002092080783116e-06,
+      "loss": 0.1934,
+      "step": 1788
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7899567015230853e-06,
+      "loss": 0.2112,
+      "step": 1789
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7797148091557247e-06,
+      "loss": 0.1948,
+      "step": 1790
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.769483548481646e-06,
+      "loss": 0.2023,
+      "step": 1791
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.759262936988083e-06,
+      "loss": 0.2026,
+      "step": 1792
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7490529921440777e-06,
+      "loss": 0.2027,
+      "step": 1793
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7388537314004394e-06,
+      "loss": 0.1991,
+      "step": 1794
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7286651721897136e-06,
+      "loss": 0.1968,
+      "step": 1795
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.718487331926157e-06,
+      "loss": 0.1917,
+      "step": 1796
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7083202280057084e-06,
+      "loss": 0.212,
+      "step": 1797
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6981638778059427e-06,
+      "loss": 0.1916,
+      "step": 1798
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.688018298686069e-06,
+      "loss": 0.2014,
+      "step": 1799
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6778835079868857e-06,
+      "loss": 0.1917,
+      "step": 1800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6677595230307393e-06,
+      "loss": 0.2062,
+      "step": 1801
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6576463611215186e-06,
+      "loss": 0.1939,
+      "step": 1802
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.647544039544615e-06,
+      "loss": 0.1832,
+      "step": 1803
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.63745257556688e-06,
+      "loss": 0.1969,
+      "step": 1804
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.627371986436621e-06,
+      "loss": 0.2014,
+      "step": 1805
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.617302289383543e-06,
+      "loss": 0.1961,
+      "step": 1806
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607243501618749e-06,
+      "loss": 0.2125,
+      "step": 1807
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.597195640334691e-06,
+      "loss": 0.2009,
+      "step": 1808
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5871587227051385e-06,
+      "loss": 0.1969,
+      "step": 1809
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.577132765885164e-06,
+      "loss": 0.2033,
+      "step": 1810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5671177870111084e-06,
+      "loss": 0.1834,
+      "step": 1811
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.557113803200537e-06,
+      "loss": 0.1813,
+      "step": 1812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.547120831552233e-06,
+      "loss": 0.2085,
+      "step": 1813
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5371388891461576e-06,
+      "loss": 0.1929,
+      "step": 1814
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.527167993043411e-06,
+      "loss": 0.2174,
+      "step": 1815
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.517208160286223e-06,
+      "loss": 0.2104,
+      "step": 1816
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5072594078979118e-06,
+      "loss": 0.2163,
+      "step": 1817
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.497321752882856e-06,
+      "loss": 0.2026,
+      "step": 1818
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4873952122264643e-06,
+      "loss": 0.2104,
+      "step": 1819
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4774798028951507e-06,
+      "loss": 0.2021,
+      "step": 1820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4675755418363054e-06,
+      "loss": 0.1846,
+      "step": 1821
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.457682445978263e-06,
+      "loss": 0.1996,
+      "step": 1822
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4478005322302766e-06,
+      "loss": 0.2034,
+      "step": 1823
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.437929817482477e-06,
+      "loss": 0.1997,
+      "step": 1824
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.428070318605865e-06,
+      "loss": 0.196,
+      "step": 1825
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4182220524522712e-06,
+      "loss": 0.2037,
+      "step": 1826
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.408385035854317e-06,
+      "loss": 0.2031,
+      "step": 1827
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3985592856254033e-06,
+      "loss": 0.2048,
+      "step": 1828
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3887448185596796e-06,
+      "loss": 0.1945,
+      "step": 1829
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.378941651431996e-06,
+      "loss": 0.2059,
+      "step": 1830
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3691498009979017e-06,
+      "loss": 0.1877,
+      "step": 1831
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359369283993602e-06,
+      "loss": 0.1885,
+      "step": 1832
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3496001171359204e-06,
+      "loss": 0.2032,
+      "step": 1833
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3398423171222928e-06,
+      "loss": 0.1939,
+      "step": 1834
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3300959006307253e-06,
+      "loss": 0.2006,
+      "step": 1835
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3203608843197575e-06,
+      "loss": 0.1922,
+      "step": 1836
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.310637284828454e-06,
+      "loss": 0.208,
+      "step": 1837
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.300925118776368e-06,
+      "loss": 0.2026,
+      "step": 1838
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2912244027634953e-06,
+      "loss": 0.1891,
+      "step": 1839
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2815351533702777e-06,
+      "loss": 0.2084,
+      "step": 1840
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.271857387157554e-06,
+      "loss": 0.2184,
+      "step": 1841
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.262191120666528e-06,
+      "loss": 0.1937,
+      "step": 1842
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2525363704187564e-06,
+      "loss": 0.1864,
+      "step": 1843
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.242893152916113e-06,
+      "loss": 0.1987,
+      "step": 1844
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.233261484640753e-06,
+      "loss": 0.2221,
+      "step": 1845
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.223641382055096e-06,
+      "loss": 0.1965,
+      "step": 1846
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214032861601796e-06,
+      "loss": 0.2008,
+      "step": 1847
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.204435939703705e-06,
+      "loss": 0.2283,
+      "step": 1848
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1948506327638596e-06,
+      "loss": 0.2124,
+      "step": 1849
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185276957165432e-06,
+      "loss": 0.2023,
+      "step": 1850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1757149292717216e-06,
+      "loss": 0.2186,
+      "step": 1851
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.166164565426123e-06,
+      "loss": 0.2195,
+      "step": 1852
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1566258819520847e-06,
+      "loss": 0.1994,
+      "step": 1853
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.147098895153098e-06,
+      "loss": 0.1934,
+      "step": 1854
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.2174,
+      "step": 1855
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1280800766942586e-06,
+      "loss": 0.2056,
+      "step": 1856
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1185882775413123e-06,
+      "loss": 0.1935,
+      "step": 1857
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.109108240077183e-06,
+      "loss": 0.1891,
+      "step": 1858
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099639980505118e-06,
+      "loss": 0.1949,
+      "step": 1859
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0901835150082417e-06,
+      "loss": 0.1922,
+      "step": 1860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0807388597495204e-06,
+      "loss": 0.217,
+      "step": 1861
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.071306030871726e-06,
+      "loss": 0.1973,
+      "step": 1862
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0618850444974237e-06,
+      "loss": 0.2073,
+      "step": 1863
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0524759167289408e-06,
+      "loss": 0.1968,
+      "step": 1864
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0430786636483223e-06,
+      "loss": 0.2012,
+      "step": 1865
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0336933013173307e-06,
+      "loss": 0.2009,
+      "step": 1866
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0243198457774025e-06,
+      "loss": 0.1981,
+      "step": 1867
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.014958313049615e-06,
+      "loss": 0.1781,
+      "step": 1868
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005608719134674e-06,
+      "loss": 0.174,
+      "step": 1869
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.99627108001288e-06,
+      "loss": 0.2055,
+      "step": 1870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9869454116440998e-06,
+      "loss": 0.2039,
+      "step": 1871
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.977631729967735e-06,
+      "loss": 0.2141,
+      "step": 1872
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9683300509027025e-06,
+      "loss": 0.1975,
+      "step": 1873
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9590403903474073e-06,
+      "loss": 0.1962,
+      "step": 1874
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.949762764179711e-06,
+      "loss": 0.206,
+      "step": 1875
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.940497188256907e-06,
+      "loss": 0.1887,
+      "step": 1876
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.931243678415683e-06,
+      "loss": 0.2054,
+      "step": 1877
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.922002250472119e-06,
+      "loss": 0.2248,
+      "step": 1878
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9127729202216303e-06,
+      "loss": 0.181,
+      "step": 1879
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9035557034389638e-06,
+      "loss": 0.2023,
+      "step": 1880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.894350615878163e-06,
+      "loss": 0.2072,
+      "step": 1881
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8851576732725307e-06,
+      "loss": 0.2104,
+      "step": 1882
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.875976891334621e-06,
+      "loss": 0.1995,
+      "step": 1883
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8668082857562006e-06,
+      "loss": 0.2004,
+      "step": 1884
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.857651872208218e-06,
+      "loss": 0.1956,
+      "step": 1885
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.848507666340794e-06,
+      "loss": 0.1727,
+      "step": 1886
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.839375683783179e-06,
+      "loss": 0.195,
+      "step": 1887
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8302559401437246e-06,
+      "loss": 0.1895,
+      "step": 1888
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.821148451009874e-06,
+      "loss": 0.2057,
+      "step": 1889
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8120532319481255e-06,
+      "loss": 0.2237,
+      "step": 1890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.802970298503992e-06,
+      "loss": 0.2111,
+      "step": 1891
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7938996662020024e-06,
+      "loss": 0.2208,
+      "step": 1892
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7848413505456564e-06,
+      "loss": 0.1899,
+      "step": 1893
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7757953670173942e-06,
+      "loss": 0.188,
+      "step": 1894
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7667617310785887e-06,
+      "loss": 0.2027,
+      "step": 1895
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7577404581695035e-06,
+      "loss": 0.2061,
+      "step": 1896
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7487315637092726e-06,
+      "loss": 0.187,
+      "step": 1897
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.739735063095874e-06,
+      "loss": 0.1969,
+      "step": 1898
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7307509717060954e-06,
+      "loss": 0.1946,
+      "step": 1899
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.721779304895521e-06,
+      "loss": 0.2113,
+      "step": 1900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7128200779984994e-06,
+      "loss": 0.1968,
+      "step": 1901
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7038733063281177e-06,
+      "loss": 0.1895,
+      "step": 1902
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6949390051761635e-06,
+      "loss": 0.1983,
+      "step": 1903
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.686017189813124e-06,
+      "loss": 0.1985,
+      "step": 1904
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6771078754881417e-06,
+      "loss": 0.1948,
+      "step": 1905
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.668211077428985e-06,
+      "loss": 0.2171,
+      "step": 1906
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.659326810842039e-06,
+      "loss": 0.2052,
+      "step": 1907
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6504550909122674e-06,
+      "loss": 0.1875,
+      "step": 1908
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641595932803183e-06,
+      "loss": 0.1993,
+      "step": 1909
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.632749351656835e-06,
+      "loss": 0.1961,
+      "step": 1910
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6239153625937786e-06,
+      "loss": 0.191,
+      "step": 1911
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.615093980713036e-06,
+      "loss": 0.2001,
+      "step": 1912
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6062852210920886e-06,
+      "loss": 0.2067,
+      "step": 1913
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.597489098786847e-06,
+      "loss": 0.2127,
+      "step": 1914
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5887056288316127e-06,
+      "loss": 0.1873,
+      "step": 1915
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.57993482623907e-06,
+      "loss": 0.2,
+      "step": 1916
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5711767060002457e-06,
+      "loss": 0.2044,
+      "step": 1917
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5624312830844944e-06,
+      "loss": 0.2008,
+      "step": 1918
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5536985724394726e-06,
+      "loss": 0.219,
+      "step": 1919
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.544978588991096e-06,
+      "loss": 0.2005,
+      "step": 1920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5362713476435385e-06,
+      "loss": 0.1888,
+      "step": 1921
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527576863279192e-06,
+      "loss": 0.1916,
+      "step": 1922
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5188951507586422e-06,
+      "loss": 0.2043,
+      "step": 1923
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.510226224920649e-06,
+      "loss": 0.2021,
+      "step": 1924
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5015701005821156e-06,
+      "loss": 0.2025,
+      "step": 1925
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.492926792538061e-06,
+      "loss": 0.2076,
+      "step": 1926
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.484296315561604e-06,
+      "loss": 0.2144,
+      "step": 1927
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4756786844039327e-06,
+      "loss": 0.218,
+      "step": 1928
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4670739137942723e-06,
+      "loss": 0.2195,
+      "step": 1929
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.458482018439874e-06,
+      "loss": 0.2021,
+      "step": 1930
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.449903013025984e-06,
+      "loss": 0.2008,
+      "step": 1931
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.441336912215807e-06,
+      "loss": 0.182,
+      "step": 1932
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4327837306504996e-06,
+      "loss": 0.1874,
+      "step": 1933
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424243482949138e-06,
+      "loss": 0.1994,
+      "step": 1934
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415716183708684e-06,
+      "loss": 0.1964,
+      "step": 1935
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4072018475039736e-06,
+      "loss": 0.2022,
+      "step": 1936
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3987004888876898e-06,
+      "loss": 0.212,
+      "step": 1937
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.390212122390323e-06,
+      "loss": 0.1851,
+      "step": 1938
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.381736762520168e-06,
+      "loss": 0.1915,
+      "step": 1939
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3732744237632888e-06,
+      "loss": 0.1993,
+      "step": 1940
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3648251205834827e-06,
+      "loss": 0.2049,
+      "step": 1941
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.356388867422278e-06,
+      "loss": 0.1921,
+      "step": 1942
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.347965678698896e-06,
+      "loss": 0.1728,
+      "step": 1943
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.1938,
+      "step": 1944
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331158552130791e-06,
+      "loss": 0.2273,
+      "step": 1945
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.322774643012766e-06,
+      "loss": 0.2163,
+      "step": 1946
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3144038557858915e-06,
+      "loss": 0.1815,
+      "step": 1947
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3060462047574973e-06,
+      "loss": 0.1973,
+      "step": 1948
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297701704212455e-06,
+      "loss": 0.1924,
+      "step": 1949
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2893703684131608e-06,
+      "loss": 0.1964,
+      "step": 1950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2810522115995126e-06,
+      "loss": 0.2047,
+      "step": 1951
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272747247988875e-06,
+      "loss": 0.2019,
+      "step": 1952
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2644554917760674e-06,
+      "loss": 0.213,
+      "step": 1953
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25617695713334e-06,
+      "loss": 0.2003,
+      "step": 1954
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2479116582103355e-06,
+      "loss": 0.2141,
+      "step": 1955
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2396596091340805e-06,
+      "loss": 0.203,
+      "step": 1956
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2314208240089553e-06,
+      "loss": 0.1886,
+      "step": 1957
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.223195316916662e-06,
+      "loss": 0.2099,
+      "step": 1958
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2149831019162173e-06,
+      "loss": 0.1841,
+      "step": 1959
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.206784193043916e-06,
+      "loss": 0.2197,
+      "step": 1960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1985986043133057e-06,
+      "loss": 0.2017,
+      "step": 1961
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190426349715171e-06,
+      "loss": 0.2001,
+      "step": 1962
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1822674432175104e-06,
+      "loss": 0.2007,
+      "step": 1963
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1741218987654977e-06,
+      "loss": 0.2072,
+      "step": 1964
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.165989730281475e-06,
+      "loss": 0.2117,
+      "step": 1965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.157870951664924e-06,
+      "loss": 0.2288,
+      "step": 1966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149765576792433e-06,
+      "loss": 0.2013,
+      "step": 1967
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.141673619517687e-06,
+      "loss": 0.1913,
+      "step": 1968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.133595093671439e-06,
+      "loss": 0.202,
+      "step": 1969
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1255300130614744e-06,
+      "loss": 0.2157,
+      "step": 1970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1174783914726106e-06,
+      "loss": 0.2074,
+      "step": 1971
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.109440242666657e-06,
+      "loss": 0.2048,
+      "step": 1972
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.101415580382389e-06,
+      "loss": 0.2136,
+      "step": 1973
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0934044183355384e-06,
+      "loss": 0.1843,
+      "step": 1974
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.085406770218761e-06,
+      "loss": 0.1893,
+      "step": 1975
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0774226497016127e-06,
+      "loss": 0.1902,
+      "step": 1976
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.069452070430529e-06,
+      "loss": 0.1986,
+      "step": 1977
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.061495046028803e-06,
+      "loss": 0.2107,
+      "step": 1978
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.053551590096553e-06,
+      "loss": 0.1828,
+      "step": 1979
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045621716210713e-06,
+      "loss": 0.1953,
+      "step": 1980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.037705437925003e-06,
+      "loss": 0.2023,
+      "step": 1981
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0298027687698974e-06,
+      "loss": 0.1865,
+      "step": 1982
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0219137222526188e-06,
+      "loss": 0.1962,
+      "step": 1983
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0140383118571028e-06,
+      "loss": 0.1985,
+      "step": 1984
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0061765510439747e-06,
+      "loss": 0.1862,
+      "step": 1985
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9983284532505343e-06,
+      "loss": 0.2033,
+      "step": 1986
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9904940318907317e-06,
+      "loss": 0.1897,
+      "step": 1987
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9826733003551302e-06,
+      "loss": 0.1977,
+      "step": 1988
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974866272010908e-06,
+      "loss": 0.1913,
+      "step": 1989
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.967072960201808e-06,
+      "loss": 0.1977,
+      "step": 1990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9592933782481404e-06,
+      "loss": 0.1895,
+      "step": 1991
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9515275394467446e-06,
+      "loss": 0.2074,
+      "step": 1992
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9437754570709675e-06,
+      "loss": 0.2012,
+      "step": 1993
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.936037144370645e-06,
+      "loss": 0.2161,
+      "step": 1994
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.928312614572083e-06,
+      "loss": 0.1978,
+      "step": 1995
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.920601880878019e-06,
+      "loss": 0.1951,
+      "step": 1996
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9129049564676203e-06,
+      "loss": 0.2103,
+      "step": 1997
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9052218544964473e-06,
+      "loss": 0.1848,
+      "step": 1998
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8975525880964375e-06,
+      "loss": 0.2053,
+      "step": 1999
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8898971703758738e-06,
+      "loss": 0.1786,
+      "step": 2000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.882255614419376e-06,
+      "loss": 0.1968,
+      "step": 2001
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8746279332878692e-06,
+      "loss": 0.212,
+      "step": 2002
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8670141400185648e-06,
+      "loss": 0.187,
+      "step": 2003
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8594142476249365e-06,
+      "loss": 0.1986,
+      "step": 2004
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.851828269096696e-06,
+      "loss": 0.1956,
+      "step": 2005
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8442562173997758e-06,
+      "loss": 0.2212,
+      "step": 2006
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8366981054763077e-06,
+      "loss": 0.1935,
+      "step": 2007
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8291539462445895e-06,
+      "loss": 0.1906,
+      "step": 2008
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8216237525990787e-06,
+      "loss": 0.1902,
+      "step": 2009
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8141075374103634e-06,
+      "loss": 0.2185,
+      "step": 2010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8066053135251305e-06,
+      "loss": 0.1775,
+      "step": 2011
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7991170937661628e-06,
+      "loss": 0.1871,
+      "step": 2012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7916428909323057e-06,
+      "loss": 0.1977,
+      "step": 2013
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7841827177984394e-06,
+      "loss": 0.2036,
+      "step": 2014
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7767365871154719e-06,
+      "loss": 0.2196,
+      "step": 2015
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7693045116103125e-06,
+      "loss": 0.2082,
+      "step": 2016
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7618865039858367e-06,
+      "loss": 0.2064,
+      "step": 2017
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.754482576920884e-06,
+      "loss": 0.1919,
+      "step": 2018
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7470927430702277e-06,
+      "loss": 0.1918,
+      "step": 2019
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7397170150645448e-06,
+      "loss": 0.1775,
+      "step": 2020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.732355405510412e-06,
+      "loss": 0.2013,
+      "step": 2021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7250079269902708e-06,
+      "loss": 0.1866,
+      "step": 2022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.717674592062407e-06,
+      "loss": 0.1982,
+      "step": 2023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7103554132609369e-06,
+      "loss": 0.1854,
+      "step": 2024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.703050403095783e-06,
+      "loss": 0.2068,
+      "step": 2025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6957595740526412e-06,
+      "loss": 0.1762,
+      "step": 2026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6884829385929779e-06,
+      "loss": 0.23,
+      "step": 2027
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6812205091539979e-06,
+      "loss": 0.2182,
+      "step": 2028
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6739722981486239e-06,
+      "loss": 0.1949,
+      "step": 2029
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6667383179654783e-06,
+      "loss": 0.2031,
+      "step": 2030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6595185809688564e-06,
+      "loss": 0.1984,
+      "step": 2031
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6523130994987125e-06,
+      "loss": 0.1929,
+      "step": 2032
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.2052,
+      "step": 2033
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6379449523758262e-06,
+      "loss": 0.1898,
+      "step": 2034
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6307823112810773e-06,
+      "loss": 0.209,
+      "step": 2035
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6236339748287565e-06,
+      "loss": 0.2027,
+      "step": 2036
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6164999552367767e-06,
+      "loss": 0.2069,
+      "step": 2037
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6093802646985834e-06,
+      "loss": 0.2043,
+      "step": 2038
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.602274915383135e-06,
+      "loss": 0.2048,
+      "step": 2039
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5951839194348684e-06,
+      "loss": 0.1879,
+      "step": 2040
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.588107288973696e-06,
+      "loss": 0.2281,
+      "step": 2041
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5810450360949758e-06,
+      "loss": 0.2021,
+      "step": 2042
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5739971728694848e-06,
+      "loss": 0.2,
+      "step": 2043
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5669637113434134e-06,
+      "loss": 0.195,
+      "step": 2044
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5599446635383341e-06,
+      "loss": 0.2079,
+      "step": 2045
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5529400414511809e-06,
+      "loss": 0.1859,
+      "step": 2046
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5459498570542343e-06,
+      "loss": 0.1848,
+      "step": 2047
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5389741222950994e-06,
+      "loss": 0.201,
+      "step": 2048
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5320128490966768e-06,
+      "loss": 0.1912,
+      "step": 2049
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5250660493571567e-06,
+      "loss": 0.2029,
+      "step": 2050
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5181337349499893e-06,
+      "loss": 0.2069,
+      "step": 2051
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5112159177238683e-06,
+      "loss": 0.1926,
+      "step": 2052
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5043126095027016e-06,
+      "loss": 0.2121,
+      "step": 2053
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4974238220856074e-06,
+      "loss": 0.1873,
+      "step": 2054
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4905495672468784e-06,
+      "loss": 0.196,
+      "step": 2055
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4836898567359737e-06,
+      "loss": 0.2089,
+      "step": 2056
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4768447022774913e-06,
+      "loss": 0.208,
+      "step": 2057
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4700141155711433e-06,
+      "loss": 0.1866,
+      "step": 2058
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4631981082917534e-06,
+      "loss": 0.1957,
+      "step": 2059
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4563966920892224e-06,
+      "loss": 0.2034,
+      "step": 2060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.449609878588506e-06,
+      "loss": 0.2083,
+      "step": 2061
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4428376793896114e-06,
+      "loss": 0.1909,
+      "step": 2062
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4360801060675578e-06,
+      "loss": 0.2095,
+      "step": 2063
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4293371701723701e-06,
+      "loss": 0.1944,
+      "step": 2064
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4226088832290575e-06,
+      "loss": 0.2194,
+      "step": 2065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.415895256737585e-06,
+      "loss": 0.1861,
+      "step": 2066
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4091963021728639e-06,
+      "loss": 0.1973,
+      "step": 2067
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4025120309847318e-06,
+      "loss": 0.192,
+      "step": 2068
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3958424545979198e-06,
+      "loss": 0.1965,
+      "step": 2069
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3891875844120517e-06,
+      "loss": 0.2026,
+      "step": 2070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.382547431801614e-06,
+      "loss": 0.1971,
+      "step": 2071
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3759220081159309e-06,
+      "loss": 0.2211,
+      "step": 2072
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.369311324679159e-06,
+      "loss": 0.1964,
+      "step": 2073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3627153927902636e-06,
+      "loss": 0.1952,
+      "step": 2074
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3561342237229858e-06,
+      "loss": 0.1888,
+      "step": 2075
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.349567828725844e-06,
+      "loss": 0.2016,
+      "step": 2076
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3430162190221007e-06,
+      "loss": 0.2072,
+      "step": 2077
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3364794058097474e-06,
+      "loss": 0.1812,
+      "step": 2078
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3299574002614901e-06,
+      "loss": 0.1954,
+      "step": 2079
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3234502135247162e-06,
+      "loss": 0.2064,
+      "step": 2080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3169578567214913e-06,
+      "loss": 0.2035,
+      "step": 2081
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3104803409485357e-06,
+      "loss": 0.1848,
+      "step": 2082
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3040176772771996e-06,
+      "loss": 0.2013,
+      "step": 2083
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.297569876753446e-06,
+      "loss": 0.1977,
+      "step": 2084
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2911369503978389e-06,
+      "loss": 0.2059,
+      "step": 2085
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.284718909205518e-06,
+      "loss": 0.21,
+      "step": 2086
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2783157641461785e-06,
+      "loss": 0.1992,
+      "step": 2087
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2719275261640584e-06,
+      "loss": 0.1989,
+      "step": 2088
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2655542061779168e-06,
+      "loss": 0.2095,
+      "step": 2089
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2591958150810101e-06,
+      "loss": 0.2068,
+      "step": 2090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.252852363741084e-06,
+      "loss": 0.1899,
+      "step": 2091
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2465238630003485e-06,
+      "loss": 0.2177,
+      "step": 2092
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2402103236754538e-06,
+      "loss": 0.1986,
+      "step": 2093
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2339117565574877e-06,
+      "loss": 0.1921,
+      "step": 2094
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.227628172411942e-06,
+      "loss": 0.2082,
+      "step": 2095
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2213595819786971e-06,
+      "loss": 0.1862,
+      "step": 2096
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2151059959720136e-06,
+      "loss": 0.2156,
+      "step": 2097
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.208867425080499e-06,
+      "loss": 0.2097,
+      "step": 2098
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2026438799671013e-06,
+      "loss": 0.2099,
+      "step": 2099
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.196435371269089e-06,
+      "loss": 0.1937,
+      "step": 2100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1902419095980221e-06,
+      "loss": 0.1784,
+      "step": 2101
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1840635055397488e-06,
+      "loss": 0.2076,
+      "step": 2102
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1779001696543802e-06,
+      "loss": 0.1994,
+      "step": 2103
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1717519124762722e-06,
+      "loss": 0.1983,
+      "step": 2104
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1656187445140066e-06,
+      "loss": 0.1982,
+      "step": 2105
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1595006762503791e-06,
+      "loss": 0.1993,
+      "step": 2106
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1533977181423706e-06,
+      "loss": 0.1929,
+      "step": 2107
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1473098806211403e-06,
+      "loss": 0.1964,
+      "step": 2108
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1412371740920036e-06,
+      "loss": 0.1806,
+      "step": 2109
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1351796089344115e-06,
+      "loss": 0.1971,
+      "step": 2110
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.129137195501936e-06,
+      "loss": 0.2062,
+      "step": 2111
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.123109944122256e-06,
+      "loss": 0.2167,
+      "step": 2112
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1170978650971276e-06,
+      "loss": 0.1956,
+      "step": 2113
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.111100968702381e-06,
+      "loss": 0.1834,
+      "step": 2114
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1051192651878938e-06,
+      "loss": 0.2048,
+      "step": 2115
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0991527647775758e-06,
+      "loss": 0.1969,
+      "step": 2116
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0932014776693512e-06,
+      "loss": 0.1913,
+      "step": 2117
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0872654140351458e-06,
+      "loss": 0.1689,
+      "step": 2118
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.081344584020858e-06,
+      "loss": 0.192,
+      "step": 2119
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0754389977463553e-06,
+      "loss": 0.1933,
+      "step": 2120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.069548665305451e-06,
+      "loss": 0.1928,
+      "step": 2121
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.2061,
+      "step": 2122
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0578138021692907e-06,
+      "loss": 0.1896,
+      "step": 2123
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.051969291531234e-06,
+      "loss": 0.1864,
+      "step": 2124
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0461400748411244e-06,
+      "loss": 0.1746,
+      "step": 2125
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.040326162062243e-06,
+      "loss": 0.1983,
+      "step": 2126
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0345275631317165e-06,
+      "loss": 0.1997,
+      "step": 2127
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0287442879604882e-06,
+      "loss": 0.2174,
+      "step": 2128
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0229763464333176e-06,
+      "loss": 0.1862,
+      "step": 2129
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0172237484087522e-06,
+      "loss": 0.1859,
+      "step": 2130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0114865037191168e-06,
+      "loss": 0.1985,
+      "step": 2131
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.005764622170493e-06,
+      "loss": 0.1843,
+      "step": 2132
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0000581135427067e-06,
+      "loss": 0.2008,
+      "step": 2133
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943669875893013e-07,
+      "loss": 0.1941,
+      "step": 2134
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.88691254037537e-07,
+      "loss": 0.1774,
+      "step": 2135
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.830309225883562e-07,
+      "loss": 0.1958,
+      "step": 2136
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77386002916383e-07,
+      "loss": 0.1969,
+      "step": 2137
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.717565046698997e-07,
+      "loss": 0.1933,
+      "step": 2138
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66142437470825e-07,
+      "loss": 0.2077,
+      "step": 2139
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.60543810914707e-07,
+      "loss": 0.1938,
+      "step": 2140
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.549606345707053e-07,
+      "loss": 0.1759,
+      "step": 2141
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.493929179815631e-07,
+      "loss": 0.2225,
+      "step": 2142
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.438406706636083e-07,
+      "loss": 0.1887,
+      "step": 2143
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.383039021067275e-07,
+      "loss": 0.1954,
+      "step": 2144
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.327826217743452e-07,
+      "loss": 0.1985,
+      "step": 2145
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.272768391034203e-07,
+      "loss": 0.2084,
+      "step": 2146
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.21786563504421e-07,
+      "loss": 0.2027,
+      "step": 2147
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.163118043613084e-07,
+      "loss": 0.204,
+      "step": 2148
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.108525710315252e-07,
+      "loss": 0.1885,
+      "step": 2149
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.054088728459786e-07,
+      "loss": 0.2101,
+      "step": 2150
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.999807191090193e-07,
+      "loss": 0.1919,
+      "step": 2151
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.945681190984334e-07,
+      "loss": 0.2048,
+      "step": 2152
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.89171082065422e-07,
+      "loss": 0.2051,
+      "step": 2153
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.837896172345827e-07,
+      "loss": 0.1976,
+      "step": 2154
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.784237338039004e-07,
+      "loss": 0.193,
+      "step": 2155
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.730734409447272e-07,
+      "loss": 0.2054,
+      "step": 2156
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.677387478017673e-07,
+      "loss": 0.1871,
+      "step": 2157
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.624196634930648e-07,
+      "loss": 0.2006,
+      "step": 2158
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.57116197109984e-07,
+      "loss": 0.1949,
+      "step": 2159
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.518283577171894e-07,
+      "loss": 0.2035,
+      "step": 2160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.465561543526435e-07,
+      "loss": 0.1683,
+      "step": 2161
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.412995960275827e-07,
+      "loss": 0.1896,
+      "step": 2162
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.360586917264979e-07,
+      "loss": 0.2021,
+      "step": 2163
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.30833450407128e-07,
+      "loss": 0.1957,
+      "step": 2164
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.256238810004425e-07,
+      "loss": 0.1965,
+      "step": 2165
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.204299924106196e-07,
+      "loss": 0.2054,
+      "step": 2166
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.152517935150384e-07,
+      "loss": 0.1957,
+      "step": 2167
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.100892931642646e-07,
+      "loss": 0.1878,
+      "step": 2168
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.049425001820255e-07,
+      "loss": 0.2027,
+      "step": 2169
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.998114233652066e-07,
+      "loss": 0.1988,
+      "step": 2170
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.946960714838269e-07,
+      "loss": 0.195,
+      "step": 2171
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.895964532810318e-07,
+      "loss": 0.2193,
+      "step": 2172
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.845125774730777e-07,
+      "loss": 0.203,
+      "step": 2173
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.794444527493039e-07,
+      "loss": 0.1886,
+      "step": 2174
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.743920877721378e-07,
+      "loss": 0.1803,
+      "step": 2175
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.693554911770695e-07,
+      "loss": 0.1724,
+      "step": 2176
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.643346715726297e-07,
+      "loss": 0.1746,
+      "step": 2177
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.593296375403914e-07,
+      "loss": 0.2047,
+      "step": 2178
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.54340397634944e-07,
+      "loss": 0.2011,
+      "step": 2179
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.493669603838827e-07,
+      "loss": 0.2142,
+      "step": 2180
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4440933428779e-07,
+      "loss": 0.1997,
+      "step": 2181
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.394675278202268e-07,
+      "loss": 0.2076,
+      "step": 2182
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.345415494277141e-07,
+      "loss": 0.1897,
+      "step": 2183
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.296314075297196e-07,
+      "loss": 0.2067,
+      "step": 2184
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.247371105186463e-07,
+      "loss": 0.1727,
+      "step": 2185
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.198586667598084e-07,
+      "loss": 0.1866,
+      "step": 2186
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.149960845914294e-07,
+      "loss": 0.2117,
+      "step": 2187
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.101493723246211e-07,
+      "loss": 0.208,
+      "step": 2188
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.053185382433691e-07,
+      "loss": 0.1995,
+      "step": 2189
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.005035906045199e-07,
+      "loss": 0.1895,
+      "step": 2190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.957045376377713e-07,
+      "loss": 0.1945,
+      "step": 2191
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.909213875456467e-07,
+      "loss": 0.1772,
+      "step": 2192
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.86154148503494e-07,
+      "loss": 0.1979,
+      "step": 2193
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.814028286594654e-07,
+      "loss": 0.2046,
+      "step": 2194
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.766674361345016e-07,
+      "loss": 0.2142,
+      "step": 2195
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.719479790223204e-07,
+      "loss": 0.1987,
+      "step": 2196
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.672444653894073e-07,
+      "loss": 0.201,
+      "step": 2197
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.625569032749912e-07,
+      "loss": 0.196,
+      "step": 2198
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.578853006910402e-07,
+      "loss": 0.1901,
+      "step": 2199
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.532296656222458e-07,
+      "loss": 0.1923,
+      "step": 2200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.485900060260042e-07,
+      "loss": 0.2081,
+      "step": 2201
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.439663298324061e-07,
+      "loss": 0.1853,
+      "step": 2202
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.393586449442302e-07,
+      "loss": 0.2025,
+      "step": 2203
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.347669592369133e-07,
+      "loss": 0.1967,
+      "step": 2204
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.30191280558553e-07,
+      "loss": 0.2025,
+      "step": 2205
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.25631616729887e-07,
+      "loss": 0.1922,
+      "step": 2206
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.210879755442789e-07,
+      "loss": 0.2091,
+      "step": 2207
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.165603647677054e-07,
+      "loss": 0.1898,
+      "step": 2208
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.120487921387463e-07,
+      "loss": 0.1933,
+      "step": 2209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.075532653685701e-07,
+      "loss": 0.2005,
+      "step": 2210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.1903,
+      "step": 2211
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.986103801120879e-07,
+      "loss": 0.201,
+      "step": 2212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.941630369109353e-07,
+      "loss": 0.1945,
+      "step": 2213
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.897317701388461e-07,
+      "loss": 0.1892,
+      "step": 2214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.853165873697253e-07,
+      "loss": 0.198,
+      "step": 2215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.809174961499931e-07,
+      "loss": 0.1618,
+      "step": 2216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.765345039985648e-07,
+      "loss": 0.189,
+      "step": 2217
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.721676184068337e-07,
+      "loss": 0.1853,
+      "step": 2218
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.678168468386691e-07,
+      "loss": 0.1847,
+      "step": 2219
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.634821967303994e-07,
+      "loss": 0.2171,
+      "step": 2220
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.591636754907937e-07,
+      "loss": 0.2092,
+      "step": 2221
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.548612905010553e-07,
+      "loss": 0.1867,
+      "step": 2222
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.505750491148121e-07,
+      "loss": 0.1908,
+      "step": 2223
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.463049586580893e-07,
+      "loss": 0.188,
+      "step": 2224
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.420510264293178e-07,
+      "loss": 0.1879,
+      "step": 2225
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.378132596993047e-07,
+      "loss": 0.1879,
+      "step": 2226
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.335916657112284e-07,
+      "loss": 0.2182,
+      "step": 2227
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.293862516806236e-07,
+      "loss": 0.2101,
+      "step": 2228
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.251970247953752e-07,
+      "loss": 0.1997,
+      "step": 2229
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.210239922156946e-07,
+      "loss": 0.2086,
+      "step": 2230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.168671610741172e-07,
+      "loss": 0.2078,
+      "step": 2231
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.127265384754865e-07,
+      "loss": 0.1681,
+      "step": 2232
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.086021314969447e-07,
+      "loss": 0.1974,
+      "step": 2233
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.04493947187914e-07,
+      "loss": 0.168,
+      "step": 2234
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.004019925700921e-07,
+      "loss": 0.1722,
+      "step": 2235
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.963262746374343e-07,
+      "loss": 0.2015,
+      "step": 2236
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.922668003561481e-07,
+      "loss": 0.2012,
+      "step": 2237
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.882235766646748e-07,
+      "loss": 0.1872,
+      "step": 2238
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.841966104736773e-07,
+      "loss": 0.2041,
+      "step": 2239
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.801859086660387e-07,
+      "loss": 0.1748,
+      "step": 2240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.761914780968369e-07,
+      "loss": 0.2005,
+      "step": 2241
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7221332559334034e-07,
+      "loss": 0.1918,
+      "step": 2242
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6825145795499684e-07,
+      "loss": 0.1818,
+      "step": 2243
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6430588195341853e-07,
+      "loss": 0.2036,
+      "step": 2244
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.60376604332371e-07,
+      "loss": 0.2018,
+      "step": 2245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5646363180776756e-07,
+      "loss": 0.1882,
+      "step": 2246
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.52566971067645e-07,
+      "loss": 0.1869,
+      "step": 2247
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4868662877216807e-07,
+      "loss": 0.1792,
+      "step": 2248
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4482261155360584e-07,
+      "loss": 0.1677,
+      "step": 2249
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.409749260163232e-07,
+      "loss": 0.2035,
+      "step": 2250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3714357873677395e-07,
+      "loss": 0.2077,
+      "step": 2251
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3332857626348866e-07,
+      "loss": 0.1944,
+      "step": 2252
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.295299251170537e-07,
+      "loss": 0.1775,
+      "step": 2253
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2574763179011434e-07,
+      "loss": 0.1934,
+      "step": 2254
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.219817027473572e-07,
+      "loss": 0.1777,
+      "step": 2255
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.182321444254944e-07,
+      "loss": 0.1898,
+      "step": 2256
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1449896323326187e-07,
+      "loss": 0.2024,
+      "step": 2257
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1078216555140103e-07,
+      "loss": 0.2083,
+      "step": 2258
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0708175773265246e-07,
+      "loss": 0.1986,
+      "step": 2259
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.033977461017424e-07,
+      "loss": 0.2063,
+      "step": 2260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9973013695537186e-07,
+      "loss": 0.2098,
+      "step": 2261
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.960789365622075e-07,
+      "loss": 0.1987,
+      "step": 2262
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9244415116287073e-07,
+      "loss": 0.1668,
+      "step": 2263
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8882578696992857e-07,
+      "loss": 0.1997,
+      "step": 2264
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.852238501678751e-07,
+      "loss": 0.2039,
+      "step": 2265
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8163834691313016e-07,
+      "loss": 0.1914,
+      "step": 2266
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.780692833340294e-07,
+      "loss": 0.2163,
+      "step": 2267
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.745166655308019e-07,
+      "loss": 0.2009,
+      "step": 2268
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.709804995755728e-07,
+      "loss": 0.2121,
+      "step": 2269
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6746079151234735e-07,
+      "loss": 0.1977,
+      "step": 2270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6395754735699896e-07,
+      "loss": 0.2022,
+      "step": 2271
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.604707730972623e-07,
+      "loss": 0.1837,
+      "step": 2272
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5700047469272467e-07,
+      "loss": 0.1899,
+      "step": 2273
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.535466580748059e-07,
+      "loss": 0.1985,
+      "step": 2274
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5010932914676164e-07,
+      "loss": 0.1997,
+      "step": 2275
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4668849378366565e-07,
+      "loss": 0.1796,
+      "step": 2276
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4328415783239646e-07,
+      "loss": 0.1773,
+      "step": 2277
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3989632711163734e-07,
+      "loss": 0.1916,
+      "step": 2278
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3652500741185754e-07,
+      "loss": 0.202,
+      "step": 2279
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3317020449530666e-07,
+      "loss": 0.196,
+      "step": 2280
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.298319240960068e-07,
+      "loss": 0.2076,
+      "step": 2281
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.265101719197328e-07,
+      "loss": 0.2001,
+      "step": 2282
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2320495364401625e-07,
+      "loss": 0.2007,
+      "step": 2283
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.199162749181273e-07,
+      "loss": 0.2148,
+      "step": 2284
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.166441413630639e-07,
+      "loss": 0.1727,
+      "step": 2285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.13388558571549e-07,
+      "loss": 0.1773,
+      "step": 2286
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.10149532108015e-07,
+      "loss": 0.2011,
+      "step": 2287
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.069270675085956e-07,
+      "loss": 0.2019,
+      "step": 2288
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0372117028111825e-07,
+      "loss": 0.1952,
+      "step": 2289
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.005318459050932e-07,
+      "loss": 0.1971,
+      "step": 2290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9735909983170443e-07,
+      "loss": 0.2042,
+      "step": 2291
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.942029374837996e-07,
+      "loss": 0.1742,
+      "step": 2292
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9106336425588356e-07,
+      "loss": 0.201,
+      "step": 2293
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.879403855141039e-07,
+      "loss": 0.211,
+      "step": 2294
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8483400659624737e-07,
+      "loss": 0.1829,
+      "step": 2295
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8174423281172923e-07,
+      "loss": 0.196,
+      "step": 2296
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.786710694415795e-07,
+      "loss": 0.1999,
+      "step": 2297
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7561452173844206e-07,
+      "loss": 0.2117,
+      "step": 2298
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.725745949265579e-07,
+      "loss": 0.1821,
+      "step": 2299
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.1885,
+      "step": 2300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.665446247314696e-07,
+      "loss": 0.2186,
+      "step": 2301
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6355459165467444e-07,
+      "loss": 0.211,
+      "step": 2302
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6058120008192855e-07,
+      "loss": 0.1945,
+      "step": 2303
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.576244550953466e-07,
+      "loss": 0.1952,
+      "step": 2304
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5468436174858833e-07,
+      "loss": 0.1745,
+      "step": 2305
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.517609250668518e-07,
+      "loss": 0.2033,
+      "step": 2306
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.488541500468666e-07,
+      "loss": 0.1863,
+      "step": 2307
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.459640416568854e-07,
+      "loss": 0.1772,
+      "step": 2308
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4309060483666993e-07,
+      "loss": 0.2138,
+      "step": 2309
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.402338444974928e-07,
+      "loss": 0.1966,
+      "step": 2310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3739376552212036e-07,
+      "loss": 0.1957,
+      "step": 2311
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.345703727648052e-07,
+      "loss": 0.1832,
+      "step": 2312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3176367105128494e-07,
+      "loss": 0.193,
+      "step": 2313
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.289736651787644e-07,
+      "loss": 0.1887,
+      "step": 2314
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2620035991591238e-07,
+      "loss": 0.1931,
+      "step": 2315
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2344376000285606e-07,
+      "loss": 0.1854,
+      "step": 2316
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2070387015116658e-07,
+      "loss": 0.2083,
+      "step": 2317
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1798069504385678e-07,
+      "loss": 0.1999,
+      "step": 2318
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1527423933536906e-07,
+      "loss": 0.1866,
+      "step": 2319
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.125845076515687e-07,
+      "loss": 0.2049,
+      "step": 2320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.099115045897382e-07,
+      "loss": 0.2011,
+      "step": 2321
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0725523471856744e-07,
+      "loss": 0.1989,
+      "step": 2322
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0461570257814368e-07,
+      "loss": 0.2148,
+      "step": 2323
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0199291267994802e-07,
+      "loss": 0.2192,
+      "step": 2324
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.993868695068457e-07,
+      "loss": 0.2006,
+      "step": 2325
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9679757751307704e-07,
+      "loss": 0.2015,
+      "step": 2326
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.94225041124253e-07,
+      "loss": 0.1972,
+      "step": 2327
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9166926473734636e-07,
+      "loss": 0.1806,
+      "step": 2328
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8913025272068063e-07,
+      "loss": 0.2012,
+      "step": 2329
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8660800941392775e-07,
+      "loss": 0.1761,
+      "step": 2330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.841025391281015e-07,
+      "loss": 0.1902,
+      "step": 2331
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8161384614554188e-07,
+      "loss": 0.1816,
+      "step": 2332
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7914193471991525e-07,
+      "loss": 0.1881,
+      "step": 2333
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.766868090762075e-07,
+      "loss": 0.1856,
+      "step": 2334
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7424847341071083e-07,
+      "loss": 0.1966,
+      "step": 2335
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.718269318910204e-07,
+      "loss": 0.1979,
+      "step": 2336
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.694221886560299e-07,
+      "loss": 0.1904,
+      "step": 2337
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6703424781591926e-07,
+      "loss": 0.1939,
+      "step": 2338
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6466311345214813e-07,
+      "loss": 0.1995,
+      "step": 2339
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.623087896174558e-07,
+      "loss": 0.2117,
+      "step": 2340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.599712803358422e-07,
+      "loss": 0.196,
+      "step": 2341
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5765058960257374e-07,
+      "loss": 0.2103,
+      "step": 2342
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.553467213841664e-07,
+      "loss": 0.1736,
+      "step": 2343
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5305967961838696e-07,
+      "loss": 0.2037,
+      "step": 2344
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5078946821423855e-07,
+      "loss": 0.2023,
+      "step": 2345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4853609105196175e-07,
+      "loss": 0.1921,
+      "step": 2346
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4629955198302016e-07,
+      "loss": 0.1904,
+      "step": 2347
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4407985483010145e-07,
+      "loss": 0.1789,
+      "step": 2348
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.418770033871053e-07,
+      "loss": 0.2009,
+      "step": 2349
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3969100141913772e-07,
+      "loss": 0.2,
+      "step": 2350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3752185266250883e-07,
+      "loss": 0.1905,
+      "step": 2351
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3536956082472074e-07,
+      "loss": 0.1944,
+      "step": 2352
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3323412958446304e-07,
+      "loss": 0.1984,
+      "step": 2353
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3111556259160941e-07,
+      "loss": 0.1967,
+      "step": 2354
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.290138634672089e-07,
+      "loss": 0.2048,
+      "step": 2355
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2692903580347915e-07,
+      "loss": 0.2129,
+      "step": 2356
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2486108316380085e-07,
+      "loss": 0.2002,
+      "step": 2357
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2281000908271336e-07,
+      "loss": 0.1851,
+      "step": 2358
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.207758170659068e-07,
+      "loss": 0.2124,
+      "step": 2359
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.187585105902156e-07,
+      "loss": 0.1951,
+      "step": 2360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1675809310361497e-07,
+      "loss": 0.1824,
+      "step": 2361
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1477456802521102e-07,
+      "loss": 0.1818,
+      "step": 2362
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1280793874523965e-07,
+      "loss": 0.195,
+      "step": 2363
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.108582086250587e-07,
+      "loss": 0.2152,
+      "step": 2364
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0892538099714023e-07,
+      "loss": 0.1833,
+      "step": 2365
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0700945916506832e-07,
+      "loss": 0.1885,
+      "step": 2366
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.051104464035313e-07,
+      "loss": 0.1859,
+      "step": 2367
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.03228345958315e-07,
+      "loss": 0.1763,
+      "step": 2368
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0136316104630062e-07,
+      "loss": 0.194,
+      "step": 2369
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.951489485545696e-08,
+      "loss": 0.2033,
+      "step": 2370
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.768355054483369e-08,
+      "loss": 0.189,
+      "step": 2371
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.586913124456255e-08,
+      "loss": 0.1825,
+      "step": 2372
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.407164005584057e-08,
+      "loss": 0.1858,
+      "step": 2373
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.229108005093467e-08,
+      "loss": 0.1949,
+      "step": 2374
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.052745427317488e-08,
+      "loss": 0.1944,
+      "step": 2375
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.878076573694328e-08,
+      "loss": 0.2056,
+      "step": 2376
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.705101742767618e-08,
+      "loss": 0.1994,
+      "step": 2377
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.533821230185313e-08,
+      "loss": 0.1812,
+      "step": 2378
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.364235328699566e-08,
+      "loss": 0.2164,
+      "step": 2379
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.196344328166184e-08,
+      "loss": 0.2036,
+      "step": 2380
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.030148515544067e-08,
+      "loss": 0.1973,
+      "step": 2381
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.865648174894325e-08,
+      "loss": 0.1916,
+      "step": 2382
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.702843587380715e-08,
+      "loss": 0.2047,
+      "step": 2383
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.541735031268316e-08,
+      "loss": 0.2258,
+      "step": 2384
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.382322781923301e-08,
+      "loss": 0.2094,
+      "step": 2385
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.224607111812609e-08,
+      "loss": 0.2122,
+      "step": 2386
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.068588290503498e-08,
+      "loss": 0.1824,
+      "step": 2387
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.914266584662988e-08,
+      "loss": 0.1738,
+      "step": 2388
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.1916,
+      "step": 2389
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.610715571550796e-08,
+      "loss": 0.1928,
+      "step": 2390
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.461486783107762e-08,
+      "loss": 0.1813,
+      "step": 2391
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.313956147789402e-08,
+      "loss": 0.2017,
+      "step": 2392
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.168123917754454e-08,
+      "loss": 0.2006,
+      "step": 2393
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.02399034225909e-08,
+      "loss": 0.2056,
+      "step": 2394
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.881555667656025e-08,
+      "loss": 0.1855,
+      "step": 2395
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.740820137394076e-08,
+      "loss": 0.1896,
+      "step": 2396
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6017839920180506e-08,
+      "loss": 0.1946,
+      "step": 2397
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4644474691680774e-08,
+      "loss": 0.1986,
+      "step": 2398
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3288108035793874e-08,
+      "loss": 0.1736,
+      "step": 2399
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1948742270817584e-08,
+      "loss": 0.2108,
+      "step": 2400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.062637968599293e-08,
+      "loss": 0.1853,
+      "step": 2401
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.93210225414964e-08,
+      "loss": 0.1905,
+      "step": 2402
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.803267306844106e-08,
+      "loss": 0.2013,
+      "step": 2403
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6761333468872125e-08,
+      "loss": 0.1872,
+      "step": 2404
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.550700591575585e-08,
+      "loss": 0.1925,
+      "step": 2405
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.426969255298841e-08,
+      "loss": 0.1976,
+      "step": 2406
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.304939549538145e-08,
+      "loss": 0.2176,
+      "step": 2407
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.184611682866213e-08,
+      "loss": 0.1989,
+      "step": 2408
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.065985860947086e-08,
+      "loss": 0.2038,
+      "step": 2409
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.949062286535799e-08,
+      "loss": 0.2035,
+      "step": 2410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8338411594777136e-08,
+      "loss": 0.1984,
+      "step": 2411
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.72032267670841e-08,
+      "loss": 0.2174,
+      "step": 2412
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.608507032253461e-08,
+      "loss": 0.1879,
+      "step": 2413
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4983944172276575e-08,
+      "loss": 0.1871,
+      "step": 2414
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.38998501983534e-08,
+      "loss": 0.2245,
+      "step": 2415
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.283279025369401e-08,
+      "loss": 0.1698,
+      "step": 2416
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1782766162115064e-08,
+      "loss": 0.1825,
+      "step": 2417
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0749779718314273e-08,
+      "loss": 0.2062,
+      "step": 2418
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9733832687869335e-08,
+      "loss": 0.212,
+      "step": 2419
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8734926807233442e-08,
+      "loss": 0.207,
+      "step": 2420
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7753063783734212e-08,
+      "loss": 0.198,
+      "step": 2421
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6788245295568117e-08,
+      "loss": 0.1937,
+      "step": 2422
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.584047299180048e-08,
+      "loss": 0.1892,
+      "step": 2423
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4909748492362162e-08,
+      "loss": 0.1951,
+      "step": 2424
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3996073388042885e-08,
+      "loss": 0.2091,
+      "step": 2425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3099449240494566e-08,
+      "loss": 0.185,
+      "step": 2426
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2219877582224657e-08,
+      "loss": 0.2125,
+      "step": 2427
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1357359916593932e-08,
+      "loss": 0.2007,
+      "step": 2428
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0511897717815367e-08,
+      "loss": 0.1906,
+      "step": 2429
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.968349243094969e-08,
+      "loss": 0.1984,
+      "step": 2430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.887214547190652e-08,
+      "loss": 0.2105,
+      "step": 2431
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8077858227436574e-08,
+      "loss": 0.2109,
+      "step": 2432
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.730063205513277e-08,
+      "loss": 0.1943,
+      "step": 2433
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6540468283429144e-08,
+      "loss": 0.1912,
+      "step": 2434
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5797368211594168e-08,
+      "loss": 0.1933,
+      "step": 2435
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5071333109732966e-08,
+      "loss": 0.1844,
+      "step": 2436
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4362364218782898e-08,
+      "loss": 0.2074,
+      "step": 2437
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3670462750511316e-08,
+      "loss": 0.1767,
+      "step": 2438
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.299562988751335e-08,
+      "loss": 0.182,
+      "step": 2439
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2337866783211916e-08,
+      "loss": 0.1987,
+      "step": 2440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1697174561854374e-08,
+      "loss": 0.2023,
+      "step": 2441
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1073554318509206e-08,
+      "loss": 0.2123,
+      "step": 2442
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0467007119067118e-08,
+      "loss": 0.1898,
+      "step": 2443
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.8775340002355e-09,
+      "loss": 0.1884,
+      "step": 2444
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.305135969541746e-09,
+      "loss": 0.2085,
+      "step": 2445
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.749814005325486e-09,
+      "loss": 0.2064,
+      "step": 2446
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.211569056741919e-09,
+      "loss": 0.1951,
+      "step": 2447
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.690402043758482e-09,
+      "loss": 0.205,
+      "step": 2448
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.186313857151517e-09,
+      "loss": 0.1785,
+      "step": 2449
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6993053585062695e-09,
+      "loss": 0.2073,
+      "step": 2450
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.229377380218005e-09,
+      "loss": 0.1776,
+      "step": 2451
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.776530725484231e-09,
+      "loss": 0.1741,
+      "step": 2452
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.340766168311362e-09,
+      "loss": 0.2039,
+      "step": 2453
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.922084453505838e-09,
+      "loss": 0.2235,
+      "step": 2454
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.520486296677451e-09,
+      "loss": 0.2032,
+      "step": 2455
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1359723842349095e-09,
+      "loss": 0.1884,
+      "step": 2456
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.768543373391387e-09,
+      "loss": 0.2091,
+      "step": 2457
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4181998921534177e-09,
+      "loss": 0.191,
+      "step": 2458
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.084942539326452e-09,
+      "loss": 0.2114,
+      "step": 2459
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7687718845148538e-09,
+      "loss": 0.1888,
+      "step": 2460
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4696884681152388e-09,
+      "loss": 0.1972,
+      "step": 2461
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1876928013209177e-09,
+      "loss": 0.195,
+      "step": 2462
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9227853661174524e-09,
+      "loss": 0.2071,
+      "step": 2463
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6749666152848787e-09,
+      "loss": 0.1935,
+      "step": 2464
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4442369723932648e-09,
+      "loss": 0.194,
+      "step": 2465
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.230596831804931e-09,
+      "loss": 0.1953,
+      "step": 2466
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0340465586733406e-09,
+      "loss": 0.2122,
+      "step": 2467
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.545864889419886e-10,
+      "loss": 0.1864,
+      "step": 2468
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.922169293421821e-10,
+      "loss": 0.1825,
+      "step": 2469
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.469381573974808e-10,
+      "loss": 0.1867,
+      "step": 2470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.187504214170357e-10,
+      "loss": 0.2052,
+      "step": 2471
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.076539404989198e-10,
+      "loss": 0.1858,
+      "step": 2472
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1364890452790777e-10,
+      "loss": 0.2015,
+      "step": 2473
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3673547417880628e-10,
+      "loss": 0.217,
+      "step": 2474
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.691378091090329e-11,
+      "loss": 0.1944,
+      "step": 2475
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4183926971298465e-11,
+      "loss": 0.2019,
+      "step": 2476
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.545985394903255e-12,
+      "loss": 0.1944,
+      "step": 2477
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.2113,
+      "step": 2478
+    },
+    {
+      "epoch": 3.0,
+      "step": 2478,
+      "total_flos": 6.32938828464862e+18,
+      "train_loss": 0.34607023420451244,
+      "train_runtime": 92308.8184,
+      "train_samples_per_second": 3.435,
+      "train_steps_per_second": 0.027
+    }
+  ],
+  "max_steps": 2478,
+  "num_train_epochs": 3,
+  "total_flos": 6.32938828464862e+18,
+  "trial_name": null,
+  "trial_params": null
+}