diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,58516 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.22460003916103247,
+  "global_step": 19500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 6.4378,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 5.9931,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 6.5129,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 6.2221,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 6.866,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 6.2606,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 5.9494,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 6.0689,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 6.0829,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 5.8833,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 5.9735,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 5.5391,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 5.6865,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 5.286,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 5.8669,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 5.4368,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 5.2647,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 5.9864,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9e-05,
+      "loss": 5.2934,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 4.8785,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1e-05,
+      "loss": 4.8154,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 4.3948,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 4.6187,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 4.2847,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 4.0182,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 4.91,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.2794,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 4.5266,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9e-05,
+      "loss": 3.6455,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 5.0947,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1e-05,
+      "loss": 5.3402,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 4.1979,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3e-05,
+      "loss": 3.6514,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.3941,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 5.016,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 3.7781,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7e-05,
+      "loss": 2.9272,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-05,
+      "loss": 5.2388,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 5.0177,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 3.0727,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1e-05,
+      "loss": 3.5789,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 4.4052,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3e-05,
+      "loss": 4.2854,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.4508,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 3.4353,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 4.8879,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7e-05,
+      "loss": 5.5409,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 4.0023,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9e-05,
+      "loss": 4.37,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 4.303,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999537165602148e-05,
+      "loss": 3.3043,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999074331204295e-05,
+      "loss": 2.4998,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998611496806443e-05,
+      "loss": 3.153,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9981486624085904e-05,
+      "loss": 1.6494,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997685828010738e-05,
+      "loss": 3.1824,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9972229936128855e-05,
+      "loss": 1.9534,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9967601592150334e-05,
+      "loss": 3.1306,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9962973248171806e-05,
+      "loss": 3.7542,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9958344904193285e-05,
+      "loss": 2.0625,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995371656021476e-05,
+      "loss": 5.8539,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9949088216236235e-05,
+      "loss": 5.0081,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994445987225771e-05,
+      "loss": 2.7199,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9939831528279186e-05,
+      "loss": 2.7836,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993520318430066e-05,
+      "loss": 4.3907,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993057484032214e-05,
+      "loss": 2.0928,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992594649634361e-05,
+      "loss": 4.4573,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992131815236509e-05,
+      "loss": 1.8654,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.991668980838656e-05,
+      "loss": 1.2354,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.991206146440804e-05,
+      "loss": 2.2481,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990743312042951e-05,
+      "loss": 3.9755,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990280477645099e-05,
+      "loss": 2.7043,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989817643247247e-05,
+      "loss": 2.062,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989354808849394e-05,
+      "loss": 1.191,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988891974451542e-05,
+      "loss": 2.0809,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988429140053689e-05,
+      "loss": 5.0056,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987966305655837e-05,
+      "loss": 2.7314,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987503471257984e-05,
+      "loss": 3.4277,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.987040636860132e-05,
+      "loss": 4.2682,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986577802462279e-05,
+      "loss": 3.6359,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986114968064427e-05,
+      "loss": 4.2905,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9856521336665744e-05,
+      "loss": 3.0396,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985189299268722e-05,
+      "loss": 2.766,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9847264648708695e-05,
+      "loss": 4.4905,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9842636304730174e-05,
+      "loss": 5.7682,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9838007960751646e-05,
+      "loss": 4.1597,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9833379616773124e-05,
+      "loss": 4.2055,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9828751272794596e-05,
+      "loss": 3.0727,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982412292881607e-05,
+      "loss": 2.741,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981949458483755e-05,
+      "loss": 2.2629,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981486624085902e-05,
+      "loss": 5.8604,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98102378968805e-05,
+      "loss": 2.8624,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.980560955290197e-05,
+      "loss": 2.5825,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.980098120892345e-05,
+      "loss": 3.3554,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.979635286494492e-05,
+      "loss": 2.8004,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.97917245209664e-05,
+      "loss": 2.6874,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.978709617698787e-05,
+      "loss": 3.5117,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.978246783300935e-05,
+      "loss": 5.0183,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.977783948903082e-05,
+      "loss": 2.3389,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.97732111450523e-05,
+      "loss": 4.539,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9768582801073774e-05,
+      "loss": 5.2634,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.976395445709525e-05,
+      "loss": 5.6091,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9759326113116724e-05,
+      "loss": 4.631,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.97546977691382e-05,
+      "loss": 3.0488,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975006942515968e-05,
+      "loss": 2.1015,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9745441081181154e-05,
+      "loss": 4.2247,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.974081273720263e-05,
+      "loss": 2.0447,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9736184393224105e-05,
+      "loss": 0.9285,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9731556049245584e-05,
+      "loss": 1.6242,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9726927705267056e-05,
+      "loss": 1.4645,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9722299361288535e-05,
+      "loss": 2.7375,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.971767101731001e-05,
+      "loss": 5.7341,
+      "step": 222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9713042673331485e-05,
+      "loss": 5.8455,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.970841432935296e-05,
+      "loss": 1.864,
+      "step": 226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9703785985374436e-05,
+      "loss": 1.9969,
+      "step": 228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.969915764139591e-05,
+      "loss": 1.8399,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.969452929741739e-05,
+      "loss": 0.997,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.968990095343886e-05,
+      "loss": 4.495,
+      "step": 234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.968527260946034e-05,
+      "loss": 1.8839,
+      "step": 236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.968064426548181e-05,
+      "loss": 2.5412,
+      "step": 238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.967601592150329e-05,
+      "loss": 2.5321,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.967138757752476e-05,
+      "loss": 2.7879,
+      "step": 242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.966675923354624e-05,
+      "loss": 4.2651,
+      "step": 244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.966213088956771e-05,
+      "loss": 1.4989,
+      "step": 246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.965750254558919e-05,
+      "loss": 1.574,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.965287420161067e-05,
+      "loss": 2.9886,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.964824585763214e-05,
+      "loss": 3.6707,
+      "step": 252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.964361751365362e-05,
+      "loss": 5.8323,
+      "step": 254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.963898916967509e-05,
+      "loss": 5.6814,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.963436082569657e-05,
+      "loss": 4.8639,
+      "step": 258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.962973248171804e-05,
+      "loss": 3.5252,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.962510413773952e-05,
+      "loss": 4.6291,
+      "step": 262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9620475793760994e-05,
+      "loss": 1.9104,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.961584744978247e-05,
+      "loss": 4.8404,
+      "step": 266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9611219105803945e-05,
+      "loss": 2.6891,
+      "step": 268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9606590761825424e-05,
+      "loss": 3.4303,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9601962417846896e-05,
+      "loss": 1.1275,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9597334073868374e-05,
+      "loss": 1.2095,
+      "step": 274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9592705729889847e-05,
+      "loss": 3.8646,
+      "step": 276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9588077385911325e-05,
+      "loss": 3.9609,
+      "step": 278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.95834490419328e-05,
+      "loss": 2.5968,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9578820697954276e-05,
+      "loss": 2.2243,
+      "step": 282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.957419235397575e-05,
+      "loss": 5.3337,
+      "step": 284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.956956400999723e-05,
+      "loss": 3.0658,
+      "step": 286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9564935666018706e-05,
+      "loss": 1.1265,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.956030732204018e-05,
+      "loss": 3.2748,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.955567897806166e-05,
+      "loss": 2.1312,
+      "step": 292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.955105063408313e-05,
+      "loss": 2.9288,
+      "step": 294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.954642229010461e-05,
+      "loss": 2.1133,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.954179394612608e-05,
+      "loss": 2.795,
+      "step": 298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.953716560214756e-05,
+      "loss": 1.094,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.953253725816903e-05,
+      "loss": 3.2405,
+      "step": 302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.952790891419051e-05,
+      "loss": 2.0241,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.952328057021198e-05,
+      "loss": 1.3409,
+      "step": 306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.951865222623346e-05,
+      "loss": 0.7884,
+      "step": 308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.951402388225493e-05,
+      "loss": 3.2695,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.950939553827641e-05,
+      "loss": 3.1318,
+      "step": 312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.950476719429788e-05,
+      "loss": 2.9392,
+      "step": 314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.950013885031936e-05,
+      "loss": 1.1243,
+      "step": 316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9495510506340834e-05,
+      "loss": 3.6301,
+      "step": 318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9490882162362306e-05,
+      "loss": 1.2792,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9486253818383785e-05,
+      "loss": 4.2146,
+      "step": 322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.948162547440526e-05,
+      "loss": 4.8716,
+      "step": 324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9476997130426736e-05,
+      "loss": 4.0848,
+      "step": 326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.947236878644821e-05,
+      "loss": 1.2625,
+      "step": 328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9467740442469686e-05,
+      "loss": 2.7873,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.946311209849116e-05,
+      "loss": 1.6087,
+      "step": 332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.945848375451264e-05,
+      "loss": 3.5022,
+      "step": 334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.945385541053411e-05,
+      "loss": 6.9392,
+      "step": 336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.944922706655559e-05,
+      "loss": 1.8005,
+      "step": 338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.944459872257706e-05,
+      "loss": 1.533,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.943997037859854e-05,
+      "loss": 3.0175,
+      "step": 342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.943534203462001e-05,
+      "loss": 1.2817,
+      "step": 344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.943071369064149e-05,
+      "loss": 2.4534,
+      "step": 346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.942608534666296e-05,
+      "loss": 2.2906,
+      "step": 348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.942145700268444e-05,
+      "loss": 1.2994,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.941682865870591e-05,
+      "loss": 3.0533,
+      "step": 352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.941220031472739e-05,
+      "loss": 3.9938,
+      "step": 354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.940757197074887e-05,
+      "loss": 3.9988,
+      "step": 356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.940294362677034e-05,
+      "loss": 1.8571,
+      "step": 358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.939831528279182e-05,
+      "loss": 1.4432,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.939368693881329e-05,
+      "loss": 3.0705,
+      "step": 362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.938905859483477e-05,
+      "loss": 1.6334,
+      "step": 364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9384430250856244e-05,
+      "loss": 3.8257,
+      "step": 366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.937980190687772e-05,
+      "loss": 2.7239,
+      "step": 368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9375173562899195e-05,
+      "loss": 0.5185,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9370545218920674e-05,
+      "loss": 3.8395,
+      "step": 372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9365916874942146e-05,
+      "loss": 5.5341,
+      "step": 374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9361288530963625e-05,
+      "loss": 4.0835,
+      "step": 376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9356660186985097e-05,
+      "loss": 5.1902,
+      "step": 378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9352031843006575e-05,
+      "loss": 5.7215,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.934740349902805e-05,
+      "loss": 4.9042,
+      "step": 382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9342775155049526e-05,
+      "loss": 5.1123,
+      "step": 384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9338146811071e-05,
+      "loss": 4.6615,
+      "step": 386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.933351846709248e-05,
+      "loss": 4.1794,
+      "step": 388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.932889012311395e-05,
+      "loss": 3.5466,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.932426177913543e-05,
+      "loss": 4.929,
+      "step": 392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.931963343515691e-05,
+      "loss": 3.9037,
+      "step": 394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.931500509117838e-05,
+      "loss": 5.3111,
+      "step": 396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.931037674719986e-05,
+      "loss": 4.9355,
+      "step": 398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.930574840322133e-05,
+      "loss": 2.9619,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.930112005924281e-05,
+      "loss": 3.3045,
+      "step": 402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.929649171526428e-05,
+      "loss": 4.3384,
+      "step": 404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.929186337128576e-05,
+      "loss": 2.4603,
+      "step": 406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.928723502730723e-05,
+      "loss": 3.3011,
+      "step": 408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.928260668332871e-05,
+      "loss": 2.0653,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.927797833935018e-05,
+      "loss": 4.6271,
+      "step": 412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.927334999537166e-05,
+      "loss": 4.0511,
+      "step": 414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.926872165139313e-05,
+      "loss": 2.9378,
+      "step": 416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.926409330741461e-05,
+      "loss": 4.206,
+      "step": 418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9259464963436084e-05,
+      "loss": 3.0345,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.925483661945756e-05,
+      "loss": 3.3985,
+      "step": 422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9250208275479035e-05,
+      "loss": 1.1632,
+      "step": 424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9245579931500514e-05,
+      "loss": 1.4889,
+      "step": 426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9240951587521986e-05,
+      "loss": 3.6674,
+      "step": 428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9236323243543464e-05,
+      "loss": 3.624,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9231694899564936e-05,
+      "loss": 3.553,
+      "step": 432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9227066555586415e-05,
+      "loss": 4.4441,
+      "step": 434
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9222438211607894e-05,
+      "loss": 2.822,
+      "step": 436
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9217809867629366e-05,
+      "loss": 2.4839,
+      "step": 438
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9213181523650845e-05,
+      "loss": 0.9495,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.920855317967232e-05,
+      "loss": 2.0021,
+      "step": 442
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9203924835693796e-05,
+      "loss": 3.4342,
+      "step": 444
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.919929649171527e-05,
+      "loss": 4.0067,
+      "step": 446
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9194668147736747e-05,
+      "loss": 6.0481,
+      "step": 448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.919003980375822e-05,
+      "loss": 3.6963,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.91854114597797e-05,
+      "loss": 3.6426,
+      "step": 452
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.918078311580117e-05,
+      "loss": 1.3831,
+      "step": 454
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.917615477182265e-05,
+      "loss": 4.9499,
+      "step": 456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.917152642784412e-05,
+      "loss": 4.6878,
+      "step": 458
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.91668980838656e-05,
+      "loss": 3.8595,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.916226973988707e-05,
+      "loss": 3.3701,
+      "step": 462
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.915764139590854e-05,
+      "loss": 3.3446,
+      "step": 464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.915301305193002e-05,
+      "loss": 2.8526,
+      "step": 466
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9148384707951494e-05,
+      "loss": 1.4705,
+      "step": 468
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.914375636397297e-05,
+      "loss": 2.4518,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9139128019994445e-05,
+      "loss": 1.545,
+      "step": 472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9134499676015924e-05,
+      "loss": 3.7903,
+      "step": 474
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9129871332037396e-05,
+      "loss": 6.0971,
+      "step": 476
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9125242988058875e-05,
+      "loss": 1.1406,
+      "step": 478
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.912061464408035e-05,
+      "loss": 1.9598,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9115986300101825e-05,
+      "loss": 2.7455,
+      "step": 482
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.91113579561233e-05,
+      "loss": 0.6213,
+      "step": 484
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9106729612144776e-05,
+      "loss": 1.0346,
+      "step": 486
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.910210126816625e-05,
+      "loss": 0.9245,
+      "step": 488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.909747292418773e-05,
+      "loss": 2.8286,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.90928445802092e-05,
+      "loss": 7.059,
+      "step": 492
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.908821623623068e-05,
+      "loss": 3.4778,
+      "step": 494
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.908358789225215e-05,
+      "loss": 3.1724,
+      "step": 496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.907895954827363e-05,
+      "loss": 4.3651,
+      "step": 498
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.90743312042951e-05,
+      "loss": 5.1223,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.906970286031658e-05,
+      "loss": 1.8804,
+      "step": 502
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.906507451633806e-05,
+      "loss": 0.892,
+      "step": 504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.906044617235953e-05,
+      "loss": 4.1566,
+      "step": 506
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.905581782838101e-05,
+      "loss": 5.5558,
+      "step": 508
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.905118948440248e-05,
+      "loss": 2.0906,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.904656114042396e-05,
+      "loss": 0.9953,
+      "step": 512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.904193279644543e-05,
+      "loss": 0.6079,
+      "step": 514
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.903730445246691e-05,
+      "loss": 1.3432,
+      "step": 516
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.903267610848838e-05,
+      "loss": 0.7198,
+      "step": 518
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.902804776450986e-05,
+      "loss": 3.1756,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9023419420531334e-05,
+      "loss": 3.1296,
+      "step": 522
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.901879107655281e-05,
+      "loss": 0.8081,
+      "step": 524
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9014162732574285e-05,
+      "loss": 0.7372,
+      "step": 526
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9009534388595764e-05,
+      "loss": 1.5253,
+      "step": 528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9004906044617236e-05,
+      "loss": 4.889,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9000277700638714e-05,
+      "loss": 8.4378,
+      "step": 532
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8995649356660186e-05,
+      "loss": 1.3812,
+      "step": 534
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8991021012681665e-05,
+      "loss": 4.1242,
+      "step": 536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.898639266870314e-05,
+      "loss": 0.7897,
+      "step": 538
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8981764324724616e-05,
+      "loss": 5.4463,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8977135980746095e-05,
+      "loss": 4.1206,
+      "step": 542
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897250763676757e-05,
+      "loss": 1.6076,
+      "step": 544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8967879292789046e-05,
+      "loss": 0.7202,
+      "step": 546
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.896325094881052e-05,
+      "loss": 1.5387,
+      "step": 548
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8958622604832e-05,
+      "loss": 0.6171,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.895399426085347e-05,
+      "loss": 2.5452,
+      "step": 552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.894936591687495e-05,
+      "loss": 1.3003,
+      "step": 554
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.894473757289642e-05,
+      "loss": 3.9205,
+      "step": 556
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.89401092289179e-05,
+      "loss": 3.4166,
+      "step": 558
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.893548088493937e-05,
+      "loss": 1.7375,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.893085254096085e-05,
+      "loss": 1.5945,
+      "step": 562
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.892622419698232e-05,
+      "loss": 5.5027,
+      "step": 564
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.89215958530038e-05,
+      "loss": 4.3663,
+      "step": 566
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.891696750902527e-05,
+      "loss": 5.005,
+      "step": 568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.891233916504675e-05,
+      "loss": 4.8787,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.890771082106822e-05,
+      "loss": 3.9977,
+      "step": 572
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.89030824770897e-05,
+      "loss": 3.4692,
+      "step": 574
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8898454133111174e-05,
+      "loss": 1.2738,
+      "step": 576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.889382578913265e-05,
+      "loss": 4.8468,
+      "step": 578
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8889197445154125e-05,
+      "loss": 2.7556,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8884569101175603e-05,
+      "loss": 0.5404,
+      "step": 582
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.887994075719708e-05,
+      "loss": 3.9631,
+      "step": 584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8875312413218554e-05,
+      "loss": 1.4846,
+      "step": 586
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.887068406924003e-05,
+      "loss": 3.8237,
+      "step": 588
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8866055725261505e-05,
+      "loss": 3.8888,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8861427381282984e-05,
+      "loss": 4.2396,
+      "step": 592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8856799037304456e-05,
+      "loss": 1.8337,
+      "step": 594
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8852170693325935e-05,
+      "loss": 4.2532,
+      "step": 596
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.884754234934741e-05,
+      "loss": 1.136,
+      "step": 598
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8842914005368886e-05,
+      "loss": 0.7694,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.883828566139036e-05,
+      "loss": 0.7106,
+      "step": 602
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8833657317411836e-05,
+      "loss": 4.6913,
+      "step": 604
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.882902897343331e-05,
+      "loss": 2.3532,
+      "step": 606
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.882440062945478e-05,
+      "loss": 4.6793,
+      "step": 608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.881977228547626e-05,
+      "loss": 1.1938,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.881514394149773e-05,
+      "loss": 0.7594,
+      "step": 612
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.881051559751921e-05,
+      "loss": 3.1088,
+      "step": 614
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.880588725354068e-05,
+      "loss": 4.2103,
+      "step": 616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.880125890956216e-05,
+      "loss": 0.5989,
+      "step": 618
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.879663056558363e-05,
+      "loss": 0.1589,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.879200222160511e-05,
+      "loss": 7.2191,
+      "step": 622
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8787373877626584e-05,
+      "loss": 4.7788,
+      "step": 624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.878274553364806e-05,
+      "loss": 2.4446,
+      "step": 626
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8778117189669535e-05,
+      "loss": 0.5752,
+      "step": 628
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8773488845691014e-05,
+      "loss": 2.4578,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8768860501712486e-05,
+      "loss": 4.306,
+      "step": 632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8764232157733964e-05,
+      "loss": 2.3142,
+      "step": 634
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8759603813755437e-05,
+      "loss": 4.3862,
+      "step": 636
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8754975469776915e-05,
+      "loss": 2.2431,
+      "step": 638
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.875034712579839e-05,
+      "loss": 3.1251,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8745718781819866e-05,
+      "loss": 4.3176,
+      "step": 642
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.874109043784134e-05,
+      "loss": 1.3775,
+      "step": 644
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.873646209386282e-05,
+      "loss": 2.3126,
+      "step": 646
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8731833749884296e-05,
+      "loss": 0.081,
+      "step": 648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.872720540590577e-05,
+      "loss": 3.7923,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.872257706192725e-05,
+      "loss": 1.8392,
+      "step": 652
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 1.9216,
+      "step": 654
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.87133203739702e-05,
+      "loss": 3.298,
+      "step": 656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.870869202999167e-05,
+      "loss": 2.7509,
+      "step": 658
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.870406368601315e-05,
+      "loss": 1.967,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.869943534203462e-05,
+      "loss": 5.5098,
+      "step": 662
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.86948069980561e-05,
+      "loss": 0.5408,
+      "step": 664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.869017865407757e-05,
+      "loss": 3.6871,
+      "step": 666
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.868555031009905e-05,
+      "loss": 2.8948,
+      "step": 668
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.868092196612052e-05,
+      "loss": 4.973,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8676293622142e-05,
+      "loss": 5.2988,
+      "step": 672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.867166527816347e-05,
+      "loss": 3.648,
+      "step": 674
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.866703693418495e-05,
+      "loss": 2.7992,
+      "step": 676
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8662408590206424e-05,
+      "loss": 2.3377,
+      "step": 678
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.86577802462279e-05,
+      "loss": 2.461,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8653151902249375e-05,
+      "loss": 4.7363,
+      "step": 682
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8648523558270853e-05,
+      "loss": 3.8594,
+      "step": 684
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8643895214292326e-05,
+      "loss": 4.734,
+      "step": 686
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8639266870313804e-05,
+      "loss": 3.6982,
+      "step": 688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.863463852633528e-05,
+      "loss": 4.6073,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8630010182356755e-05,
+      "loss": 3.7078,
+      "step": 692
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8625381838378234e-05,
+      "loss": 3.6916,
+      "step": 694
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8620753494399706e-05,
+      "loss": 5.4478,
+      "step": 696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8616125150421185e-05,
+      "loss": 3.211,
+      "step": 698
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.861149680644266e-05,
+      "loss": 2.4355,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8606868462464136e-05,
+      "loss": 1.4907,
+      "step": 702
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.860224011848561e-05,
+      "loss": 1.1325,
+      "step": 704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8597611774507087e-05,
+      "loss": 1.8143,
+      "step": 706
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.859298343052856e-05,
+      "loss": 0.7965,
+      "step": 708
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.858835508655004e-05,
+      "loss": 2.4407,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.858372674257151e-05,
+      "loss": 4.7933,
+      "step": 712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857909839859299e-05,
+      "loss": 0.0581,
+      "step": 714
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.857447005461446e-05,
+      "loss": 2.051,
+      "step": 716
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.856984171063594e-05,
+      "loss": 2.0997,
+      "step": 718
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.856521336665741e-05,
+      "loss": 0.146,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.856058502267889e-05,
+      "loss": 0.2953,
+      "step": 722
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.855595667870036e-05,
+      "loss": 3.3334,
+      "step": 724
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.855132833472184e-05,
+      "loss": 0.3533,
+      "step": 726
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.854669999074332e-05,
+      "loss": 0.0405,
+      "step": 728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.854207164676479e-05,
+      "loss": 3.2629,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.853744330278627e-05,
+      "loss": 6.1213,
+      "step": 732
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.853281495880774e-05,
+      "loss": 7.558,
+      "step": 734
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.852818661482922e-05,
+      "loss": 5.9161,
+      "step": 736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.852355827085069e-05,
+      "loss": 4.3886,
+      "step": 738
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.851892992687217e-05,
+      "loss": 4.0207,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8514301582893644e-05,
+      "loss": 3.3258,
+      "step": 742
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.850967323891512e-05,
+      "loss": 2.4253,
+      "step": 744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8505044894936595e-05,
+      "loss": 3.8694,
+      "step": 746
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8500416550958074e-05,
+      "loss": 3.3424,
+      "step": 748
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8495788206979546e-05,
+      "loss": 2.9284,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8491159863001025e-05,
+      "loss": 1.9573,
+      "step": 752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84865315190225e-05,
+      "loss": 3.541,
+      "step": 754
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.848190317504397e-05,
+      "loss": 4.2104,
+      "step": 756
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.847727483106545e-05,
+      "loss": 2.616,
+      "step": 758
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.847264648708692e-05,
+      "loss": 2.9753,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84680181431084e-05,
+      "loss": 3.9563,
+      "step": 762
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.846338979912987e-05,
+      "loss": 1.3559,
+      "step": 764
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845876145515135e-05,
+      "loss": 3.4534,
+      "step": 766
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845413311117282e-05,
+      "loss": 3.7294,
+      "step": 768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84495047671943e-05,
+      "loss": 4.3839,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.844487642321577e-05,
+      "loss": 0.8009,
+      "step": 772
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.844024807923725e-05,
+      "loss": 2.3561,
+      "step": 774
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.843561973525872e-05,
+      "loss": 4.2417,
+      "step": 776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84309913912802e-05,
+      "loss": 2.0557,
+      "step": 778
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8426363047301674e-05,
+      "loss": 2.45,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.842173470332315e-05,
+      "loss": 2.6521,
+      "step": 782
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8417106359344625e-05,
+      "loss": 1.4045,
+      "step": 784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8412478015366104e-05,
+      "loss": 1.2441,
+      "step": 786
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8407849671387576e-05,
+      "loss": 4.9924,
+      "step": 788
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8403221327409054e-05,
+      "loss": 4.7287,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8398592983430526e-05,
+      "loss": 4.5657,
+      "step": 792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8393964639452005e-05,
+      "loss": 0.8063,
+      "step": 794
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8389336295473484e-05,
+      "loss": 1.8242,
+      "step": 796
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8384707951494956e-05,
+      "loss": 3.2677,
+      "step": 798
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8380079607516435e-05,
+      "loss": 0.5575,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.837545126353791e-05,
+      "loss": 2.2497,
+      "step": 802
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8370822919559386e-05,
+      "loss": 6.6717,
+      "step": 804
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.836619457558086e-05,
+      "loss": 0.9386,
+      "step": 806
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8361566231602337e-05,
+      "loss": 1.7609,
+      "step": 808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.835693788762381e-05,
+      "loss": 4.1894,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.835230954364529e-05,
+      "loss": 1.7989,
+      "step": 812
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834768119966676e-05,
+      "loss": 3.2573,
+      "step": 814
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.834305285568824e-05,
+      "loss": 2.7204,
+      "step": 816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833842451170971e-05,
+      "loss": 1.7687,
+      "step": 818
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833379616773119e-05,
+      "loss": 2.5711,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832916782375266e-05,
+      "loss": 1.7332,
+      "step": 822
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.832453947977414e-05,
+      "loss": 2.1463,
+      "step": 824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.831991113579561e-05,
+      "loss": 2.1947,
+      "step": 826
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.831528279181709e-05,
+      "loss": 0.4706,
+      "step": 828
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.831065444783856e-05,
+      "loss": 3.9636,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.830602610386004e-05,
+      "loss": 0.602,
+      "step": 832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8301397759881514e-05,
+      "loss": 1.0117,
+      "step": 834
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829676941590299e-05,
+      "loss": 1.9096,
+      "step": 836
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829214107192447e-05,
+      "loss": 1.9683,
+      "step": 838
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828751272794594e-05,
+      "loss": 1.4493,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.828288438396742e-05,
+      "loss": 4.6948,
+      "step": 842
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8278256039988894e-05,
+      "loss": 3.917,
+      "step": 844
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.827362769601037e-05,
+      "loss": 2.5498,
+      "step": 846
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8268999352031845e-05,
+      "loss": 1.7735,
+      "step": 848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8264371008053324e-05,
+      "loss": 3.3149,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8259742664074796e-05,
+      "loss": 4.0292,
+      "step": 852
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8255114320096275e-05,
+      "loss": 1.8101,
+      "step": 854
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.825048597611775e-05,
+      "loss": 0.3017,
+      "step": 856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8245857632139226e-05,
+      "loss": 1.8147,
+      "step": 858
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82412292881607e-05,
+      "loss": 2.2203,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8236600944182176e-05,
+      "loss": 5.7717,
+      "step": 862
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.823197260020365e-05,
+      "loss": 2.463,
+      "step": 864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.822734425622513e-05,
+      "loss": 3.7682,
+      "step": 866
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.82227159122466e-05,
+      "loss": 3.6976,
+      "step": 868
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.821808756826808e-05,
+      "loss": 1.8342,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.821345922428955e-05,
+      "loss": 2.4817,
+      "step": 872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.820883088031103e-05,
+      "loss": 2.0065,
+      "step": 874
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.820420253633251e-05,
+      "loss": 1.5848,
+      "step": 876
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.819957419235398e-05,
+      "loss": 2.1813,
+      "step": 878
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.819494584837546e-05,
+      "loss": 1.1161,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.819031750439693e-05,
+      "loss": 2.0882,
+      "step": 882
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.818568916041841e-05,
+      "loss": 3.7063,
+      "step": 884
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.818106081643988e-05,
+      "loss": 6.5713,
+      "step": 886
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.817643247246136e-05,
+      "loss": 1.6113,
+      "step": 888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.817180412848283e-05,
+      "loss": 1.4084,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.816717578450431e-05,
+      "loss": 0.8458,
+      "step": 892
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.816254744052578e-05,
+      "loss": 1.0353,
+      "step": 894
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.815791909654726e-05,
+      "loss": 1.0656,
+      "step": 896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.815329075256873e-05,
+      "loss": 4.2288,
+      "step": 898
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8148662408590206e-05,
+      "loss": 4.648,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8144034064611685e-05,
+      "loss": 1.6827,
+      "step": 902
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.813940572063316e-05,
+      "loss": 0.1,
+      "step": 904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8134777376654636e-05,
+      "loss": 2.2172,
+      "step": 906
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.813014903267611e-05,
+      "loss": 4.0769,
+      "step": 908
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.812552068869759e-05,
+      "loss": 2.6073,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.812089234471906e-05,
+      "loss": 1.4241,
+      "step": 912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.811626400074054e-05,
+      "loss": 1.6268,
+      "step": 914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.811163565676201e-05,
+      "loss": 1.5726,
+      "step": 916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.810700731278349e-05,
+      "loss": 2.3525,
+      "step": 918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.810237896880496e-05,
+      "loss": 2.3828,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.809775062482644e-05,
+      "loss": 6.4977,
+      "step": 922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.809312228084791e-05,
+      "loss": 6.0585,
+      "step": 924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.808849393686939e-05,
+      "loss": 5.0878,
+      "step": 926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.808386559289086e-05,
+      "loss": 4.847,
+      "step": 928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807923724891234e-05,
+      "loss": 3.2003,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807460890493381e-05,
+      "loss": 2.2836,
+      "step": 932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.806998056095529e-05,
+      "loss": 3.1416,
+      "step": 934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8065352216976764e-05,
+      "loss": 2.0173,
+      "step": 936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.806072387299824e-05,
+      "loss": 3.2826,
+      "step": 938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8056095529019715e-05,
+      "loss": 2.8288,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8051467185041193e-05,
+      "loss": 2.2719,
+      "step": 942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.804683884106267e-05,
+      "loss": 1.6449,
+      "step": 944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8042210497084144e-05,
+      "loss": 2.3678,
+      "step": 946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.803758215310562e-05,
+      "loss": 2.3104,
+      "step": 948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8032953809127095e-05,
+      "loss": 0.9652,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8028325465148574e-05,
+      "loss": 6.1099,
+      "step": 952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8023697121170046e-05,
+      "loss": 0.4649,
+      "step": 954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8019068777191525e-05,
+      "loss": 2.009,
+      "step": 956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8014440433213e-05,
+      "loss": 2.3097,
+      "step": 958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8009812089234476e-05,
+      "loss": 0.0982,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800518374525595e-05,
+      "loss": 1.0323,
+      "step": 962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8000555401277426e-05,
+      "loss": 3.1926,
+      "step": 964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.79959270572989e-05,
+      "loss": 2.0062,
+      "step": 966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.799129871332038e-05,
+      "loss": 2.9063,
+      "step": 968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.798667036934185e-05,
+      "loss": 2.1665,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.798204202536333e-05,
+      "loss": 0.1262,
+      "step": 972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.79774136813848e-05,
+      "loss": 5.0226,
+      "step": 974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.797278533740628e-05,
+      "loss": 1.4879,
+      "step": 976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.796815699342775e-05,
+      "loss": 4.3901,
+      "step": 978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.796352864944923e-05,
+      "loss": 2.0089,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.795890030547071e-05,
+      "loss": 2.4167,
+      "step": 982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.795427196149218e-05,
+      "loss": 2.0942,
+      "step": 984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.794964361751366e-05,
+      "loss": 2.2375,
+      "step": 986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.794501527353513e-05,
+      "loss": 1.3551,
+      "step": 988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.794038692955661e-05,
+      "loss": 3.0897,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.793575858557808e-05,
+      "loss": 3.2452,
+      "step": 992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.793113024159956e-05,
+      "loss": 0.1318,
+      "step": 994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.792650189762103e-05,
+      "loss": 3.9303,
+      "step": 996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.792187355364251e-05,
+      "loss": 6.3771,
+      "step": 998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7917245209663984e-05,
+      "loss": 5.8938,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.791261686568546e-05,
+      "loss": 2.6714,
+      "step": 1002
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7907988521706935e-05,
+      "loss": 1.3251,
+      "step": 1004
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7903360177728414e-05,
+      "loss": 7.1614,
+      "step": 1006
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7898731833749886e-05,
+      "loss": 2.2772,
+      "step": 1008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7894103489771365e-05,
+      "loss": 1.617,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788947514579284e-05,
+      "loss": 4.3641,
+      "step": 1012
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7884846801814315e-05,
+      "loss": 1.5456,
+      "step": 1014
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788021845783579e-05,
+      "loss": 4.6484,
+      "step": 1016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7875590113857266e-05,
+      "loss": 4.0756,
+      "step": 1018
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.787096176987874e-05,
+      "loss": 1.2109,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.786633342590022e-05,
+      "loss": 1.7558,
+      "step": 1022
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7861705081921696e-05,
+      "loss": 2.5459,
+      "step": 1024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.785707673794317e-05,
+      "loss": 3.3481,
+      "step": 1026
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.785244839396465e-05,
+      "loss": 3.3966,
+      "step": 1028
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.784782004998612e-05,
+      "loss": 2.0624,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.78431917060076e-05,
+      "loss": 0.0244,
+      "step": 1032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.783856336202907e-05,
+      "loss": 0.8578,
+      "step": 1034
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.783393501805055e-05,
+      "loss": 2.8285,
+      "step": 1036
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782930667407202e-05,
+      "loss": 1.2024,
+      "step": 1038
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.78246783300935e-05,
+      "loss": 0.2802,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.782004998611497e-05,
+      "loss": 2.2741,
+      "step": 1042
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7815421642136443e-05,
+      "loss": 0.4118,
+      "step": 1044
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7810793298157915e-05,
+      "loss": 2.8413,
+      "step": 1046
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7806164954179394e-05,
+      "loss": 1.1633,
+      "step": 1048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.780153661020087e-05,
+      "loss": 3.6136,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7796908266222345e-05,
+      "loss": 0.0636,
+      "step": 1052
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7792279922243824e-05,
+      "loss": 1.746,
+      "step": 1054
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7787651578265296e-05,
+      "loss": 3.7166,
+      "step": 1056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7783023234286775e-05,
+      "loss": 4.5057,
+      "step": 1058
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.777839489030825e-05,
+      "loss": 2.6392,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7773766546329726e-05,
+      "loss": 2.7932,
+      "step": 1062
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.77691382023512e-05,
+      "loss": 4.3139,
+      "step": 1064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7764509858372677e-05,
+      "loss": 3.9198,
+      "step": 1066
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.775988151439415e-05,
+      "loss": 0.9431,
+      "step": 1068
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.775525317041563e-05,
+      "loss": 3.6874,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.77506248264371e-05,
+      "loss": 1.9133,
+      "step": 1072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.774599648245858e-05,
+      "loss": 1.7569,
+      "step": 1074
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.774136813848005e-05,
+      "loss": 5.1998,
+      "step": 1076
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.773673979450153e-05,
+      "loss": 0.2728,
+      "step": 1078
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7732111450523e-05,
+      "loss": 4.4559,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.772748310654448e-05,
+      "loss": 2.8929,
+      "step": 1082
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.772285476256595e-05,
+      "loss": 0.0211,
+      "step": 1084
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.771822641858743e-05,
+      "loss": 4.8431,
+      "step": 1086
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.771359807460891e-05,
+      "loss": 0.0501,
+      "step": 1088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.770896973063038e-05,
+      "loss": 6.2431,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.770434138665186e-05,
+      "loss": 1.4295,
+      "step": 1092
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.769971304267333e-05,
+      "loss": 0.0235,
+      "step": 1094
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.769508469869481e-05,
+      "loss": 2.5061,
+      "step": 1096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.769045635471628e-05,
+      "loss": 2.8102,
+      "step": 1098
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.768582801073776e-05,
+      "loss": 10.0612,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7681199666759234e-05,
+      "loss": 0.1956,
+      "step": 1102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.767657132278071e-05,
+      "loss": 1.5776,
+      "step": 1104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7671942978802185e-05,
+      "loss": 2.259,
+      "step": 1106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7667314634823664e-05,
+      "loss": 1.4943,
+      "step": 1108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7662686290845136e-05,
+      "loss": 4.6757,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7658057946866615e-05,
+      "loss": 3.1171,
+      "step": 1112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.765342960288809e-05,
+      "loss": 3.6605,
+      "step": 1114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7648801258909566e-05,
+      "loss": 3.1163,
+      "step": 1116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.764417291493104e-05,
+      "loss": 1.498,
+      "step": 1118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7639544570952516e-05,
+      "loss": 1.723,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.763491622697399e-05,
+      "loss": 2.3644,
+      "step": 1122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.763028788299547e-05,
+      "loss": 2.5378,
+      "step": 1124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.762565953901694e-05,
+      "loss": 3.1757,
+      "step": 1126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.762103119503842e-05,
+      "loss": 3.868,
+      "step": 1128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.76164028510599e-05,
+      "loss": 3.4017,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.761177450708137e-05,
+      "loss": 0.5591,
+      "step": 1132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.760714616310285e-05,
+      "loss": 0.0602,
+      "step": 1134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.760251781912432e-05,
+      "loss": 2.1817,
+      "step": 1136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75978894751458e-05,
+      "loss": 1.0818,
+      "step": 1138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.759326113116727e-05,
+      "loss": 2.8268,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.758863278718875e-05,
+      "loss": 0.0036,
+      "step": 1142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.758400444321022e-05,
+      "loss": 0.088,
+      "step": 1144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75793760992317e-05,
+      "loss": 0.8022,
+      "step": 1146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.757474775525317e-05,
+      "loss": 1.0228,
+      "step": 1148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.757011941127465e-05,
+      "loss": 0.7927,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.756549106729612e-05,
+      "loss": 0.0035,
+      "step": 1152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75608627233176e-05,
+      "loss": 0.0874,
+      "step": 1154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7556234379339074e-05,
+      "loss": 0.0305,
+      "step": 1156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.755160603536055e-05,
+      "loss": 6.2571,
+      "step": 1158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7546977691382025e-05,
+      "loss": 8.5162,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7542349347403504e-05,
+      "loss": 12.1018,
+      "step": 1162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7537721003424976e-05,
+      "loss": 3.5762,
+      "step": 1164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7533092659446455e-05,
+      "loss": 2.0671,
+      "step": 1166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7528464315467927e-05,
+      "loss": 3.1923,
+      "step": 1168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7523835971489405e-05,
+      "loss": 3.4033,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7519207627510884e-05,
+      "loss": 3.8199,
+      "step": 1172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7514579283532356e-05,
+      "loss": 3.4156,
+      "step": 1174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7509950939553835e-05,
+      "loss": 2.3559,
+      "step": 1176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.750532259557531e-05,
+      "loss": 4.286,
+      "step": 1178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7500694251596786e-05,
+      "loss": 3.4982,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.749606590761826e-05,
+      "loss": 2.1,
+      "step": 1182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.749143756363974e-05,
+      "loss": 5.3904,
+      "step": 1184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748680921966121e-05,
+      "loss": 0.4751,
+      "step": 1186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748218087568268e-05,
+      "loss": 3.8687,
+      "step": 1188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.747755253170415e-05,
+      "loss": 0.5556,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.747292418772563e-05,
+      "loss": 3.1869,
+      "step": 1192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7468295843747104e-05,
+      "loss": 5.0351,
+      "step": 1194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.746366749976858e-05,
+      "loss": 2.7529,
+      "step": 1196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.745903915579006e-05,
+      "loss": 2.7592,
+      "step": 1198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.745441081181153e-05,
+      "loss": 4.916,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.744978246783301e-05,
+      "loss": 1.9435,
+      "step": 1202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7445154123854484e-05,
+      "loss": 1.746,
+      "step": 1204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.744052577987596e-05,
+      "loss": 3.8881,
+      "step": 1206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7435897435897435e-05,
+      "loss": 2.4149,
+      "step": 1208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7431269091918914e-05,
+      "loss": 3.1034,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7426640747940386e-05,
+      "loss": 4.1012,
+      "step": 1212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7422012403961865e-05,
+      "loss": 2.249,
+      "step": 1214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.741738405998334e-05,
+      "loss": 5.4898,
+      "step": 1216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7412755716004816e-05,
+      "loss": 3.3004,
+      "step": 1218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.740812737202629e-05,
+      "loss": 4.122,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7403499028047766e-05,
+      "loss": 5.1891,
+      "step": 1222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.739887068406924e-05,
+      "loss": 2.6417,
+      "step": 1224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.739424234009072e-05,
+      "loss": 1.9748,
+      "step": 1226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.738961399611219e-05,
+      "loss": 3.2174,
+      "step": 1228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.738498565213367e-05,
+      "loss": 4.7975,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.738035730815514e-05,
+      "loss": 1.3937,
+      "step": 1232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.737572896417662e-05,
+      "loss": 0.4165,
+      "step": 1234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.73711006201981e-05,
+      "loss": 3.279,
+      "step": 1236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736647227621957e-05,
+      "loss": 1.9009,
+      "step": 1238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736184393224105e-05,
+      "loss": 3.4443,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.735721558826252e-05,
+      "loss": 5.6324,
+      "step": 1242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7352587244284e-05,
+      "loss": 3.6197,
+      "step": 1244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.734795890030547e-05,
+      "loss": 5.076,
+      "step": 1246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.734333055632695e-05,
+      "loss": 0.6866,
+      "step": 1248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.733870221234842e-05,
+      "loss": 4.533,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.73340738683699e-05,
+      "loss": 1.2522,
+      "step": 1252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.732944552439137e-05,
+      "loss": 3.1234,
+      "step": 1254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.732481718041285e-05,
+      "loss": 2.1485,
+      "step": 1256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7320188836434324e-05,
+      "loss": 1.3479,
+      "step": 1258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.73155604924558e-05,
+      "loss": 0.9807,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7310932148477275e-05,
+      "loss": 2.0259,
+      "step": 1262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7306303804498754e-05,
+      "loss": 3.2539,
+      "step": 1264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7301675460520226e-05,
+      "loss": 1.5512,
+      "step": 1266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7297047116541705e-05,
+      "loss": 0.6629,
+      "step": 1268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7292418772563177e-05,
+      "loss": 0.94,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7287790428584655e-05,
+      "loss": 1.3583,
+      "step": 1272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.728316208460613e-05,
+      "loss": 0.0072,
+      "step": 1274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7278533740627606e-05,
+      "loss": 3.2477,
+      "step": 1276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7273905396649085e-05,
+      "loss": 4.2692,
+      "step": 1278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.726927705267056e-05,
+      "loss": 4.784,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7264648708692036e-05,
+      "loss": 2.0589,
+      "step": 1282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.726002036471351e-05,
+      "loss": 0.7848,
+      "step": 1284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.725539202073499e-05,
+      "loss": 1.1568,
+      "step": 1286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.725076367675646e-05,
+      "loss": 5.0926,
+      "step": 1288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.724613533277794e-05,
+      "loss": 2.0791,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.724150698879941e-05,
+      "loss": 3.311,
+      "step": 1292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.723687864482089e-05,
+      "loss": 3.7369,
+      "step": 1294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.723225030084236e-05,
+      "loss": 2.28,
+      "step": 1296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722762195686384e-05,
+      "loss": 0.9142,
+      "step": 1298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722299361288531e-05,
+      "loss": 4.9183,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.721836526890679e-05,
+      "loss": 3.4118,
+      "step": 1302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.721373692492826e-05,
+      "loss": 4.4652,
+      "step": 1304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.720910858094974e-05,
+      "loss": 1.2692,
+      "step": 1306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.720448023697121e-05,
+      "loss": 2.6757,
+      "step": 1308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.719985189299269e-05,
+      "loss": 2.8063,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7195223549014164e-05,
+      "loss": 3.1885,
+      "step": 1312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.719059520503564e-05,
+      "loss": 0.8567,
+      "step": 1314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.718596686105712e-05,
+      "loss": 3.2438,
+      "step": 1316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7181338517078594e-05,
+      "loss": 1.719,
+      "step": 1318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.717671017310007e-05,
+      "loss": 1.287,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7172081829121544e-05,
+      "loss": 1.4392,
+      "step": 1322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.716745348514302e-05,
+      "loss": 5.0394,
+      "step": 1324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7162825141164495e-05,
+      "loss": 3.4047,
+      "step": 1326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7158196797185974e-05,
+      "loss": 1.6015,
+      "step": 1328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7153568453207446e-05,
+      "loss": 3.368,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7148940109228925e-05,
+      "loss": 1.796,
+      "step": 1332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.714431176525039e-05,
+      "loss": 0.9212,
+      "step": 1334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.713968342127187e-05,
+      "loss": 0.4147,
+      "step": 1336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.713505507729334e-05,
+      "loss": 0.2373,
+      "step": 1338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.713042673331482e-05,
+      "loss": 3.3293,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.71257983893363e-05,
+      "loss": 8.4659,
+      "step": 1342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.712117004535777e-05,
+      "loss": 1.597,
+      "step": 1344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.711654170137925e-05,
+      "loss": 1.5281,
+      "step": 1346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.711191335740072e-05,
+      "loss": 0.0873,
+      "step": 1348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.71072850134222e-05,
+      "loss": 2.4538,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.710265666944367e-05,
+      "loss": 3.7778,
+      "step": 1352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.709802832546515e-05,
+      "loss": 3.5011,
+      "step": 1354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.709339998148662e-05,
+      "loss": 2.4013,
+      "step": 1356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.70887716375081e-05,
+      "loss": 3.8523,
+      "step": 1358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7084143293529574e-05,
+      "loss": 5.9043,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.707951494955105e-05,
+      "loss": 3.6095,
+      "step": 1362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7074886605572525e-05,
+      "loss": 2.0603,
+      "step": 1364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7070258261594004e-05,
+      "loss": 1.9184,
+      "step": 1366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7065629917615476e-05,
+      "loss": 2.148,
+      "step": 1368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7061001573636955e-05,
+      "loss": 4.1726,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.705637322965843e-05,
+      "loss": 2.5698,
+      "step": 1372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7051744885679905e-05,
+      "loss": 2.2628,
+      "step": 1374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.704711654170138e-05,
+      "loss": 0.1632,
+      "step": 1376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7042488197722856e-05,
+      "loss": 4.9712,
+      "step": 1378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.703785985374433e-05,
+      "loss": 2.634,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.703323150976581e-05,
+      "loss": 1.1508,
+      "step": 1382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7028603165787286e-05,
+      "loss": 6.1856,
+      "step": 1384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.702397482180876e-05,
+      "loss": 1.4108,
+      "step": 1386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.701934647783024e-05,
+      "loss": 3.7405,
+      "step": 1388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.701471813385171e-05,
+      "loss": 0.0315,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.701008978987319e-05,
+      "loss": 4.4584,
+      "step": 1392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.700546144589466e-05,
+      "loss": 3.6878,
+      "step": 1394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.700083310191614e-05,
+      "loss": 2.3816,
+      "step": 1396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.699620475793761e-05,
+      "loss": 2.7244,
+      "step": 1398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.699157641395909e-05,
+      "loss": 0.7517,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.698694806998056e-05,
+      "loss": 2.4456,
+      "step": 1402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.698231972600204e-05,
+      "loss": 2.1693,
+      "step": 1404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.697769138202351e-05,
+      "loss": 5.5787,
+      "step": 1406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.697306303804499e-05,
+      "loss": 3.1504,
+      "step": 1408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.696843469406646e-05,
+      "loss": 0.8895,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.696380635008794e-05,
+      "loss": 3.5219,
+      "step": 1412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6959178006109414e-05,
+      "loss": 1.7531,
+      "step": 1414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.695454966213089e-05,
+      "loss": 4.6263,
+      "step": 1416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6949921318152365e-05,
+      "loss": 6.6263,
+      "step": 1418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6945292974173844e-05,
+      "loss": 2.2338,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6940664630195316e-05,
+      "loss": 1.1218,
+      "step": 1422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6936036286216794e-05,
+      "loss": 2.904,
+      "step": 1424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.693140794223827e-05,
+      "loss": 2.2043,
+      "step": 1426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6926779598259745e-05,
+      "loss": 1.9168,
+      "step": 1428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6922151254281224e-05,
+      "loss": 4.4268,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6917522910302696e-05,
+      "loss": 2.6503,
+      "step": 1432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6912894566324175e-05,
+      "loss": 1.7103,
+      "step": 1434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.690826622234565e-05,
+      "loss": 0.2614,
+      "step": 1436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6903637878367126e-05,
+      "loss": 4.9934,
+      "step": 1438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.68990095343886e-05,
+      "loss": 2.4072,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.689438119041008e-05,
+      "loss": 4.9945,
+      "step": 1442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.688975284643155e-05,
+      "loss": 6.048,
+      "step": 1444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.688512450245303e-05,
+      "loss": 1.2272,
+      "step": 1446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.68804961584745e-05,
+      "loss": 5.4235,
+      "step": 1448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.687586781449598e-05,
+      "loss": 3.3449,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.687123947051745e-05,
+      "loss": 3.3197,
+      "step": 1452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.686661112653893e-05,
+      "loss": 1.8447,
+      "step": 1454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.68619827825604e-05,
+      "loss": 2.2925,
+      "step": 1456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.685735443858188e-05,
+      "loss": 3.6103,
+      "step": 1458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.685272609460335e-05,
+      "loss": 0.566,
+      "step": 1460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.684809775062483e-05,
+      "loss": 4.8005,
+      "step": 1462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.684346940664631e-05,
+      "loss": 2.8564,
+      "step": 1464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.683884106266778e-05,
+      "loss": 3.8502,
+      "step": 1466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.683421271868926e-05,
+      "loss": 2.3621,
+      "step": 1468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.682958437471073e-05,
+      "loss": 1.4551,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.682495603073221e-05,
+      "loss": 2.2992,
+      "step": 1472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6820327686753683e-05,
+      "loss": 3.1056,
+      "step": 1474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.681569934277516e-05,
+      "loss": 2.9353,
+      "step": 1476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.681107099879663e-05,
+      "loss": 3.2521,
+      "step": 1478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6806442654818106e-05,
+      "loss": 1.2604,
+      "step": 1480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.680181431083958e-05,
+      "loss": 3.3738,
+      "step": 1482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.679718596686106e-05,
+      "loss": 0.6927,
+      "step": 1484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.679255762288253e-05,
+      "loss": 2.8132,
+      "step": 1486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.678792927890401e-05,
+      "loss": 1.5098,
+      "step": 1488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.678330093492549e-05,
+      "loss": 3.0815,
+      "step": 1490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.677867259094696e-05,
+      "loss": 0.0186,
+      "step": 1492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.677404424696844e-05,
+      "loss": 2.8723,
+      "step": 1494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.676941590298991e-05,
+      "loss": 0.7,
+      "step": 1496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.676478755901139e-05,
+      "loss": 0.0015,
+      "step": 1498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.676015921503286e-05,
+      "loss": 8.2901,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.675553087105434e-05,
+      "loss": 3.3564,
+      "step": 1502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.675090252707581e-05,
+      "loss": 4.3119,
+      "step": 1504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674627418309729e-05,
+      "loss": 1.7273,
+      "step": 1506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674164583911876e-05,
+      "loss": 6.9064,
+      "step": 1508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673701749514024e-05,
+      "loss": 1.4491,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673238915116171e-05,
+      "loss": 2.1862,
+      "step": 1512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.672776080718319e-05,
+      "loss": 2.4411,
+      "step": 1514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6723132463204664e-05,
+      "loss": 1.8262,
+      "step": 1516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.671850411922614e-05,
+      "loss": 9.0044,
+      "step": 1518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6713875775247615e-05,
+      "loss": 2.2658,
+      "step": 1520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6709247431269094e-05,
+      "loss": 4.8632,
+      "step": 1522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6704619087290566e-05,
+      "loss": 1.8087,
+      "step": 1524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6699990743312044e-05,
+      "loss": 3.459,
+      "step": 1526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6695362399333517e-05,
+      "loss": 1.8268,
+      "step": 1528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6690734055354995e-05,
+      "loss": 0.1453,
+      "step": 1530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6686105711376474e-05,
+      "loss": 1.4866,
+      "step": 1532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6681477367397946e-05,
+      "loss": 1.0276,
+      "step": 1534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6676849023419425e-05,
+      "loss": 4.1424,
+      "step": 1536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66722206794409e-05,
+      "loss": 10.1489,
+      "step": 1538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6667592335462376e-05,
+      "loss": 4.4767,
+      "step": 1540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666296399148385e-05,
+      "loss": 2.2751,
+      "step": 1542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665833564750533e-05,
+      "loss": 2.2312,
+      "step": 1544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66537073035268e-05,
+      "loss": 5.567,
+      "step": 1546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.664907895954828e-05,
+      "loss": 2.2757,
+      "step": 1548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.664445061556975e-05,
+      "loss": 2.6249,
+      "step": 1550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663982227159123e-05,
+      "loss": 2.2077,
+      "step": 1552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66351939276127e-05,
+      "loss": 0.018,
+      "step": 1554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663056558363418e-05,
+      "loss": 2.4558,
+      "step": 1556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662593723965565e-05,
+      "loss": 1.0583,
+      "step": 1558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662130889567713e-05,
+      "loss": 1.729,
+      "step": 1560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66166805516986e-05,
+      "loss": 1.484,
+      "step": 1562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.661205220772008e-05,
+      "loss": 3.7384,
+      "step": 1564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660742386374155e-05,
+      "loss": 6.3257,
+      "step": 1566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.660279551976303e-05,
+      "loss": 3.1436,
+      "step": 1568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.659816717578451e-05,
+      "loss": 0.9981,
+      "step": 1570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.659353883180598e-05,
+      "loss": 2.2064,
+      "step": 1572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.658891048782746e-05,
+      "loss": 3.3594,
+      "step": 1574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6584282143848933e-05,
+      "loss": 1.7759,
+      "step": 1576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657965379987041e-05,
+      "loss": 1.3102,
+      "step": 1578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6575025455891884e-05,
+      "loss": 3.0085,
+      "step": 1580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.657039711191336e-05,
+      "loss": 1.6332,
+      "step": 1582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6565768767934835e-05,
+      "loss": 1.7029,
+      "step": 1584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6561140423956314e-05,
+      "loss": 0.1442,
+      "step": 1586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6556512079977786e-05,
+      "loss": 2.6524,
+      "step": 1588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6551883735999265e-05,
+      "loss": 0.8463,
+      "step": 1590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.654725539202074e-05,
+      "loss": 0.0362,
+      "step": 1592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6542627048042216e-05,
+      "loss": 2.4726,
+      "step": 1594
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.653799870406369e-05,
+      "loss": 2.7328,
+      "step": 1596
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6533370360085167e-05,
+      "loss": 0.2323,
+      "step": 1598
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.652874201610664e-05,
+      "loss": 0.1138,
+      "step": 1600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.652411367212812e-05,
+      "loss": 5.126,
+      "step": 1602
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.651948532814959e-05,
+      "loss": 3.0895,
+      "step": 1604
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.651485698417107e-05,
+      "loss": 2.0055,
+      "step": 1606
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.651022864019254e-05,
+      "loss": 1.3187,
+      "step": 1608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.650560029621402e-05,
+      "loss": 2.1532,
+      "step": 1610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.65009719522355e-05,
+      "loss": 0.2399,
+      "step": 1612
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.649634360825697e-05,
+      "loss": 9.1947,
+      "step": 1614
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.649171526427845e-05,
+      "loss": 4.7747,
+      "step": 1616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.648708692029992e-05,
+      "loss": 5.3616,
+      "step": 1618
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64824585763214e-05,
+      "loss": 4.9785,
+      "step": 1620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.647783023234287e-05,
+      "loss": 2.9585,
+      "step": 1622
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6473201888364344e-05,
+      "loss": 3.2767,
+      "step": 1624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6468573544385816e-05,
+      "loss": 4.0326,
+      "step": 1626
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6463945200407295e-05,
+      "loss": 1.9793,
+      "step": 1628
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6459316856428767e-05,
+      "loss": 1.3678,
+      "step": 1630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6454688512450245e-05,
+      "loss": 2.1165,
+      "step": 1632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.645006016847172e-05,
+      "loss": 2.348,
+      "step": 1634
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6445431824493196e-05,
+      "loss": 1.477,
+      "step": 1636
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6440803480514675e-05,
+      "loss": 1.371,
+      "step": 1638
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.643617513653615e-05,
+      "loss": 3.6031,
+      "step": 1640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6431546792557626e-05,
+      "loss": 3.5409,
+      "step": 1642
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64269184485791e-05,
+      "loss": 2.3525,
+      "step": 1644
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642229010460058e-05,
+      "loss": 1.8493,
+      "step": 1646
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.641766176062205e-05,
+      "loss": 5.3567,
+      "step": 1648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.641303341664353e-05,
+      "loss": 5.4484,
+      "step": 1650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6408405072665e-05,
+      "loss": 0.9547,
+      "step": 1652
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.640377672868648e-05,
+      "loss": 4.832,
+      "step": 1654
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639914838470795e-05,
+      "loss": 3.5683,
+      "step": 1656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639452004072943e-05,
+      "loss": 2.5118,
+      "step": 1658
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63898916967509e-05,
+      "loss": 4.6071,
+      "step": 1660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.638526335277238e-05,
+      "loss": 4.4391,
+      "step": 1662
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.638063500879385e-05,
+      "loss": 2.2106,
+      "step": 1664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637600666481533e-05,
+      "loss": 0.224,
+      "step": 1666
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63713783208368e-05,
+      "loss": 1.7189,
+      "step": 1668
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636674997685828e-05,
+      "loss": 1.6031,
+      "step": 1670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6362121632879754e-05,
+      "loss": 2.0248,
+      "step": 1672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.635749328890123e-05,
+      "loss": 2.8778,
+      "step": 1674
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.635286494492271e-05,
+      "loss": 2.194,
+      "step": 1676
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6348236600944184e-05,
+      "loss": 7.6742,
+      "step": 1678
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.634360825696566e-05,
+      "loss": 2.3626,
+      "step": 1680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6338979912987134e-05,
+      "loss": 0.3874,
+      "step": 1682
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.633435156900861e-05,
+      "loss": 0.5834,
+      "step": 1684
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6329723225030085e-05,
+      "loss": 2.0645,
+      "step": 1686
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6325094881051564e-05,
+      "loss": 5.1255,
+      "step": 1688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6320466537073036e-05,
+      "loss": 2.7201,
+      "step": 1690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6315838193094515e-05,
+      "loss": 0.5743,
+      "step": 1692
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631120984911599e-05,
+      "loss": 0.6093,
+      "step": 1694
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6306581505137466e-05,
+      "loss": 1.1557,
+      "step": 1696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.630195316115894e-05,
+      "loss": 6.0955,
+      "step": 1698
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6297324817180417e-05,
+      "loss": 2.3893,
+      "step": 1700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.629269647320189e-05,
+      "loss": 1.033,
+      "step": 1702
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628806812922337e-05,
+      "loss": 3.5923,
+      "step": 1704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628343978524484e-05,
+      "loss": 1.1247,
+      "step": 1706
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.627881144126632e-05,
+      "loss": 3.5498,
+      "step": 1708
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.627418309728779e-05,
+      "loss": 1.8788,
+      "step": 1710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626955475330927e-05,
+      "loss": 4.0412,
+      "step": 1712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626492640933074e-05,
+      "loss": 0.8718,
+      "step": 1714
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626029806535222e-05,
+      "loss": 2.8557,
+      "step": 1716
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.62556697213737e-05,
+      "loss": 1.8011,
+      "step": 1718
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.625104137739517e-05,
+      "loss": 0.5937,
+      "step": 1720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.624641303341665e-05,
+      "loss": 5.9732,
+      "step": 1722
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.624178468943812e-05,
+      "loss": 7.3994,
+      "step": 1724
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.62371563454596e-05,
+      "loss": 6.1346,
+      "step": 1726
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.623252800148107e-05,
+      "loss": 0.8119,
+      "step": 1728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.622789965750255e-05,
+      "loss": 2.5117,
+      "step": 1730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.622327131352402e-05,
+      "loss": 4.7919,
+      "step": 1732
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.62186429695455e-05,
+      "loss": 3.0197,
+      "step": 1734
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6214014625566974e-05,
+      "loss": 2.4058,
+      "step": 1736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.620938628158845e-05,
+      "loss": 2.6577,
+      "step": 1738
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6204757937609925e-05,
+      "loss": 3.2956,
+      "step": 1740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6200129593631404e-05,
+      "loss": 2.5949,
+      "step": 1742
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6195501249652876e-05,
+      "loss": 1.5059,
+      "step": 1744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6190872905674355e-05,
+      "loss": 0.5341,
+      "step": 1746
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.618624456169583e-05,
+      "loss": 0.5147,
+      "step": 1748
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6181616217717306e-05,
+      "loss": 1.8775,
+      "step": 1750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.617698787373878e-05,
+      "loss": 0.1786,
+      "step": 1752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6172359529760256e-05,
+      "loss": 7.1488,
+      "step": 1754
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.616773118578173e-05,
+      "loss": 0.0758,
+      "step": 1756
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.616310284180321e-05,
+      "loss": 1.1736,
+      "step": 1758
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6158474497824686e-05,
+      "loss": 4.8311,
+      "step": 1760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 4.3117,
+      "step": 1762
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.614921780986764e-05,
+      "loss": 4.9715,
+      "step": 1764
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.614458946588911e-05,
+      "loss": 12.1501,
+      "step": 1766
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.613996112191058e-05,
+      "loss": 7.2045,
+      "step": 1768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.613533277793205e-05,
+      "loss": 0.2252,
+      "step": 1770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.613070443395353e-05,
+      "loss": 1.8361,
+      "step": 1772
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6126076089975004e-05,
+      "loss": 6.1041,
+      "step": 1774
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.612144774599648e-05,
+      "loss": 0.4606,
+      "step": 1776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6116819402017955e-05,
+      "loss": 1.9522,
+      "step": 1778
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6112191058039434e-05,
+      "loss": 3.2616,
+      "step": 1780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6107562714060906e-05,
+      "loss": 1.4915,
+      "step": 1782
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6102934370082384e-05,
+      "loss": 4.3302,
+      "step": 1784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.609830602610386e-05,
+      "loss": 2.0909,
+      "step": 1786
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6093677682125335e-05,
+      "loss": 1.996,
+      "step": 1788
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6089049338146814e-05,
+      "loss": 2.7939,
+      "step": 1790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6084420994168286e-05,
+      "loss": 1.2666,
+      "step": 1792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6079792650189765e-05,
+      "loss": 0.714,
+      "step": 1794
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.607516430621124e-05,
+      "loss": 1.7849,
+      "step": 1796
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6070535962232716e-05,
+      "loss": 3.3108,
+      "step": 1798
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.606590761825419e-05,
+      "loss": 0.7047,
+      "step": 1800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.606127927427567e-05,
+      "loss": 0.8227,
+      "step": 1802
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.605665093029714e-05,
+      "loss": 1.642,
+      "step": 1804
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.605202258631862e-05,
+      "loss": 2.8321,
+      "step": 1806
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.604739424234009e-05,
+      "loss": 0.0617,
+      "step": 1808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.604276589836157e-05,
+      "loss": 0.6797,
+      "step": 1810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.603813755438304e-05,
+      "loss": 7.7426,
+      "step": 1812
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.603350921040452e-05,
+      "loss": 0.4267,
+      "step": 1814
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.602888086642599e-05,
+      "loss": 0.2307,
+      "step": 1816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.602425252244747e-05,
+      "loss": 3.2965,
+      "step": 1818
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.601962417846894e-05,
+      "loss": 2.933,
+      "step": 1820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.601499583449042e-05,
+      "loss": 2.5302,
+      "step": 1822
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.60103674905119e-05,
+      "loss": 0.9394,
+      "step": 1824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600573914653337e-05,
+      "loss": 3.3788,
+      "step": 1826
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600111080255485e-05,
+      "loss": 0.4418,
+      "step": 1828
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.599648245857632e-05,
+      "loss": 2.1213,
+      "step": 1830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.59918541145978e-05,
+      "loss": 2.0603,
+      "step": 1832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5987225770619273e-05,
+      "loss": 4.9091,
+      "step": 1834
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.598259742664075e-05,
+      "loss": 4.5502,
+      "step": 1836
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5977969082662224e-05,
+      "loss": 5.3509,
+      "step": 1838
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.59733407386837e-05,
+      "loss": 3.1513,
+      "step": 1840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5968712394705175e-05,
+      "loss": 1.9933,
+      "step": 1842
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5964084050726654e-05,
+      "loss": 1.0844,
+      "step": 1844
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5959455706748126e-05,
+      "loss": 3.9193,
+      "step": 1846
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5954827362769605e-05,
+      "loss": 3.5403,
+      "step": 1848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.595019901879108e-05,
+      "loss": 1.5624,
+      "step": 1850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5945570674812556e-05,
+      "loss": 0.5245,
+      "step": 1852
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.594094233083403e-05,
+      "loss": 0.934,
+      "step": 1854
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5936313986855506e-05,
+      "loss": 2.399,
+      "step": 1856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.593168564287698e-05,
+      "loss": 2.5662,
+      "step": 1858
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.592705729889846e-05,
+      "loss": 2.4856,
+      "step": 1860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.592242895491993e-05,
+      "loss": 2.0161,
+      "step": 1862
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.591780061094141e-05,
+      "loss": 3.732,
+      "step": 1864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.591317226696289e-05,
+      "loss": 2.559,
+      "step": 1866
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.590854392298436e-05,
+      "loss": 0.8488,
+      "step": 1868
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.590391557900584e-05,
+      "loss": 4.8079,
+      "step": 1870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.589928723502731e-05,
+      "loss": 1.2993,
+      "step": 1872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.589465889104879e-05,
+      "loss": 3.6165,
+      "step": 1874
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.589003054707026e-05,
+      "loss": 3.3347,
+      "step": 1876
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.588540220309174e-05,
+      "loss": 1.3434,
+      "step": 1878
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.588077385911321e-05,
+      "loss": 2.7971,
+      "step": 1880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.587614551513469e-05,
+      "loss": 2.1208,
+      "step": 1882
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.587151717115616e-05,
+      "loss": 3.4964,
+      "step": 1884
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.586688882717764e-05,
+      "loss": 0.8696,
+      "step": 1886
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.586226048319911e-05,
+      "loss": 5.2081,
+      "step": 1888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.585763213922059e-05,
+      "loss": 3.3185,
+      "step": 1890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5853003795242064e-05,
+      "loss": 3.5607,
+      "step": 1892
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.584837545126354e-05,
+      "loss": 1.3931,
+      "step": 1894
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5843747107285015e-05,
+      "loss": 3.5017,
+      "step": 1896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5839118763306494e-05,
+      "loss": 1.6544,
+      "step": 1898
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5834490419327966e-05,
+      "loss": 0.6147,
+      "step": 1900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5829862075349445e-05,
+      "loss": 0.3586,
+      "step": 1902
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5825233731370923e-05,
+      "loss": 4.1892,
+      "step": 1904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5820605387392395e-05,
+      "loss": 1.0249,
+      "step": 1906
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5815977043413874e-05,
+      "loss": 5.8179,
+      "step": 1908
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5811348699435346e-05,
+      "loss": 1.4441,
+      "step": 1910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5806720355456825e-05,
+      "loss": 6.212,
+      "step": 1912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.580209201147829e-05,
+      "loss": 4.0223,
+      "step": 1914
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.579746366749977e-05,
+      "loss": 1.4515,
+      "step": 1916
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.579283532352124e-05,
+      "loss": 1.4379,
+      "step": 1918
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.578820697954272e-05,
+      "loss": 0.7094,
+      "step": 1920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.578357863556419e-05,
+      "loss": 2.5687,
+      "step": 1922
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.577895029158567e-05,
+      "loss": 0.9674,
+      "step": 1924
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.577432194760714e-05,
+      "loss": 3.2022,
+      "step": 1926
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.576969360362862e-05,
+      "loss": 0.9043,
+      "step": 1928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.57650652596501e-05,
+      "loss": 3.1792,
+      "step": 1930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.576043691567157e-05,
+      "loss": 3.1869,
+      "step": 1932
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.575580857169305e-05,
+      "loss": 0.2195,
+      "step": 1934
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5751180227714523e-05,
+      "loss": 3.6047,
+      "step": 1936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5746551883736e-05,
+      "loss": 8.8579,
+      "step": 1938
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5741923539757474e-05,
+      "loss": 2.4881,
+      "step": 1940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.573729519577895e-05,
+      "loss": 0.8224,
+      "step": 1942
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5732666851800425e-05,
+      "loss": 1.2833,
+      "step": 1944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5728038507821904e-05,
+      "loss": 0.8703,
+      "step": 1946
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5723410163843376e-05,
+      "loss": 3.5144,
+      "step": 1948
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5718781819864855e-05,
+      "loss": 1.7912,
+      "step": 1950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.571415347588633e-05,
+      "loss": 1.215,
+      "step": 1952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5709525131907806e-05,
+      "loss": 3.878,
+      "step": 1954
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.570489678792928e-05,
+      "loss": 1.3974,
+      "step": 1956
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5700268443950757e-05,
+      "loss": 2.7886,
+      "step": 1958
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.569564009997223e-05,
+      "loss": 2.0902,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.569101175599371e-05,
+      "loss": 0.813,
+      "step": 1962
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.568638341201518e-05,
+      "loss": 3.0979,
+      "step": 1964
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.568175506803666e-05,
+      "loss": 1.8209,
+      "step": 1966
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.567712672405813e-05,
+      "loss": 1.1106,
+      "step": 1968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.567249838007961e-05,
+      "loss": 2.2329,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.566787003610109e-05,
+      "loss": 0.053,
+      "step": 1972
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.566324169212256e-05,
+      "loss": 2.2613,
+      "step": 1974
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.565861334814404e-05,
+      "loss": 5.4557,
+      "step": 1976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.565398500416551e-05,
+      "loss": 1.3582,
+      "step": 1978
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.564935666018699e-05,
+      "loss": 3.0152,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.564472831620846e-05,
+      "loss": 1.0431,
+      "step": 1982
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.564009997222994e-05,
+      "loss": 0.0089,
+      "step": 1984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.563547162825141e-05,
+      "loss": 0.5129,
+      "step": 1986
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.563084328427289e-05,
+      "loss": 3.2135,
+      "step": 1988
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.562621494029436e-05,
+      "loss": 7.0357,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.562158659631584e-05,
+      "loss": 1.9964,
+      "step": 1992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5616958252337314e-05,
+      "loss": 2.6189,
+      "step": 1994
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.561232990835879e-05,
+      "loss": 7.0998,
+      "step": 1996
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5607701564380265e-05,
+      "loss": 3.1208,
+      "step": 1998
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5603073220401744e-05,
+      "loss": 4.3489,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5598444876423216e-05,
+      "loss": 3.6178,
+      "step": 2002
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5593816532444695e-05,
+      "loss": 3.1075,
+      "step": 2004
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.558918818846617e-05,
+      "loss": 1.6052,
+      "step": 2006
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5584559844487646e-05,
+      "loss": 2.2888,
+      "step": 2008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5579931500509124e-05,
+      "loss": 1.3035,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5575303156530596e-05,
+      "loss": 3.3716,
+      "step": 2012
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5570674812552075e-05,
+      "loss": 2.2311,
+      "step": 2014
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.556604646857355e-05,
+      "loss": 1.8558,
+      "step": 2016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5561418124595026e-05,
+      "loss": 3.0461,
+      "step": 2018
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.55567897806165e-05,
+      "loss": 0.5723,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.555216143663798e-05,
+      "loss": 4.4988,
+      "step": 2022
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.554753309265945e-05,
+      "loss": 3.8209,
+      "step": 2024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.554290474868093e-05,
+      "loss": 0.6857,
+      "step": 2026
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.55382764047024e-05,
+      "loss": 2.3959,
+      "step": 2028
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.553364806072388e-05,
+      "loss": 1.2031,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.552901971674535e-05,
+      "loss": 5.2028,
+      "step": 2032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.552439137276683e-05,
+      "loss": 6.1027,
+      "step": 2034
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.55197630287883e-05,
+      "loss": 2.4259,
+      "step": 2036
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.551513468480978e-05,
+      "loss": 2.304,
+      "step": 2038
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.551050634083125e-05,
+      "loss": 4.2057,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.550587799685273e-05,
+      "loss": 2.4561,
+      "step": 2042
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.55012496528742e-05,
+      "loss": 1.5298,
+      "step": 2044
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.549662130889568e-05,
+      "loss": 1.0376,
+      "step": 2046
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5491992964917154e-05,
+      "loss": 2.0048,
+      "step": 2048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.548736462093863e-05,
+      "loss": 1.3363,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.548273627696011e-05,
+      "loss": 2.1149,
+      "step": 2052
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5478107932981584e-05,
+      "loss": 2.7407,
+      "step": 2054
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.547347958900306e-05,
+      "loss": 0.7472,
+      "step": 2056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.546885124502453e-05,
+      "loss": 1.87,
+      "step": 2058
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5464222901046007e-05,
+      "loss": 1.9965,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.545959455706748e-05,
+      "loss": 4.991,
+      "step": 2062
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.545496621308896e-05,
+      "loss": 0.1791,
+      "step": 2064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.545033786911043e-05,
+      "loss": 1.6174,
+      "step": 2066
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.544570952513191e-05,
+      "loss": 1.8906,
+      "step": 2068
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.544108118115338e-05,
+      "loss": 3.155,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.543645283717486e-05,
+      "loss": 2.1355,
+      "step": 2072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.543182449319633e-05,
+      "loss": 7.8725,
+      "step": 2074
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.542719614921781e-05,
+      "loss": 1.0543,
+      "step": 2076
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.542256780523929e-05,
+      "loss": 3.4159,
+      "step": 2078
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.541793946126076e-05,
+      "loss": 3.5965,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.541331111728224e-05,
+      "loss": 1.3608,
+      "step": 2082
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.540868277330371e-05,
+      "loss": 0.6067,
+      "step": 2084
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.540405442932519e-05,
+      "loss": 2.863,
+      "step": 2086
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.539942608534666e-05,
+      "loss": 3.5894,
+      "step": 2088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.539479774136814e-05,
+      "loss": 0.1556,
+      "step": 2090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.539016939738961e-05,
+      "loss": 2.7656,
+      "step": 2092
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.538554105341109e-05,
+      "loss": 0.5127,
+      "step": 2094
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5380912709432564e-05,
+      "loss": 0.8523,
+      "step": 2096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.537628436545404e-05,
+      "loss": 2.0607,
+      "step": 2098
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5371656021475515e-05,
+      "loss": 0.6225,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5367027677496994e-05,
+      "loss": 0.1571,
+      "step": 2102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5362399333518466e-05,
+      "loss": 1.9484,
+      "step": 2104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5357770989539945e-05,
+      "loss": 0.6029,
+      "step": 2106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.535314264556142e-05,
+      "loss": 0.9368,
+      "step": 2108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5348514301582896e-05,
+      "loss": 2.6047,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.534388595760437e-05,
+      "loss": 0.0311,
+      "step": 2112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5339257613625846e-05,
+      "loss": 8.9976,
+      "step": 2114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.533462926964732e-05,
+      "loss": 4.9767,
+      "step": 2116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.53300009256688e-05,
+      "loss": 2.0942,
+      "step": 2118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5325372581690276e-05,
+      "loss": 5.3527,
+      "step": 2120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.532074423771175e-05,
+      "loss": 3.7434,
+      "step": 2122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.531611589373323e-05,
+      "loss": 4.6058,
+      "step": 2124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.53114875497547e-05,
+      "loss": 3.7413,
+      "step": 2126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.530685920577618e-05,
+      "loss": 2.5465,
+      "step": 2128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.530223086179765e-05,
+      "loss": 1.6612,
+      "step": 2130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.529760251781913e-05,
+      "loss": 1.9171,
+      "step": 2132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.52929741738406e-05,
+      "loss": 1.7376,
+      "step": 2134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528834582986208e-05,
+      "loss": 1.8551,
+      "step": 2136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528371748588355e-05,
+      "loss": 2.7661,
+      "step": 2138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.527908914190503e-05,
+      "loss": 1.8831,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.52744607979265e-05,
+      "loss": 1.327,
+      "step": 2142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.526983245394798e-05,
+      "loss": 0.8619,
+      "step": 2144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.526520410996945e-05,
+      "loss": 0.3338,
+      "step": 2146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.526057576599093e-05,
+      "loss": 5.1896,
+      "step": 2148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5255947422012404e-05,
+      "loss": 0.0308,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.525131907803388e-05,
+      "loss": 3.9201,
+      "step": 2152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5246690734055355e-05,
+      "loss": 10.1902,
+      "step": 2154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5242062390076834e-05,
+      "loss": 2.434,
+      "step": 2156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.523743404609831e-05,
+      "loss": 1.6084,
+      "step": 2158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5232805702119785e-05,
+      "loss": 2.6577,
+      "step": 2160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.522817735814126e-05,
+      "loss": 0.3432,
+      "step": 2162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5223549014162735e-05,
+      "loss": 3.3654,
+      "step": 2164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5218920670184214e-05,
+      "loss": 1.7249,
+      "step": 2166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5214292326205686e-05,
+      "loss": 4.2702,
+      "step": 2168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5209663982227165e-05,
+      "loss": 0.0735,
+      "step": 2170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520503563824864e-05,
+      "loss": 1.9043,
+      "step": 2172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5200407294270116e-05,
+      "loss": 3.6567,
+      "step": 2174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.519577895029159e-05,
+      "loss": 3.6733,
+      "step": 2176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.519115060631307e-05,
+      "loss": 5.7071,
+      "step": 2178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.518652226233454e-05,
+      "loss": 1.0858,
+      "step": 2180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.518189391835602e-05,
+      "loss": 1.3381,
+      "step": 2182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.517726557437749e-05,
+      "loss": 4.0219,
+      "step": 2184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.517263723039897e-05,
+      "loss": 1.66,
+      "step": 2186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.516800888642044e-05,
+      "loss": 2.7369,
+      "step": 2188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.516338054244192e-05,
+      "loss": 1.1413,
+      "step": 2190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.515875219846339e-05,
+      "loss": 3.1562,
+      "step": 2192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.515412385448487e-05,
+      "loss": 2.3145,
+      "step": 2194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.514949551050634e-05,
+      "loss": 2.579,
+      "step": 2196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.514486716652782e-05,
+      "loss": 1.3117,
+      "step": 2198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.51402388225493e-05,
+      "loss": 0.0994,
+      "step": 2200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.513561047857077e-05,
+      "loss": 4.6938,
+      "step": 2202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5130982134592244e-05,
+      "loss": 1.992,
+      "step": 2204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5126353790613716e-05,
+      "loss": 4.6339,
+      "step": 2206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5121725446635195e-05,
+      "loss": 0.0708,
+      "step": 2208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.511709710265667e-05,
+      "loss": 1.9962,
+      "step": 2210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5112468758678146e-05,
+      "loss": 3.7955,
+      "step": 2212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.510784041469962e-05,
+      "loss": 5.114,
+      "step": 2214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5103212070721096e-05,
+      "loss": 0.6122,
+      "step": 2216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.509858372674257e-05,
+      "loss": 1.6119,
+      "step": 2218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.509395538276405e-05,
+      "loss": 1.6431,
+      "step": 2220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.508932703878552e-05,
+      "loss": 2.0628,
+      "step": 2222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5084698694807e-05,
+      "loss": 5.4893,
+      "step": 2224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.508007035082848e-05,
+      "loss": 1.8499,
+      "step": 2226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.507544200684995e-05,
+      "loss": 0.0013,
+      "step": 2228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.507081366287143e-05,
+      "loss": 4.5373,
+      "step": 2230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.50661853188929e-05,
+      "loss": 2.4563,
+      "step": 2232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.506155697491438e-05,
+      "loss": 2.305,
+      "step": 2234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.505692863093585e-05,
+      "loss": 0.4411,
+      "step": 2236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.505230028695733e-05,
+      "loss": 7.7629,
+      "step": 2238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.50476719429788e-05,
+      "loss": 3.6271,
+      "step": 2240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.504304359900028e-05,
+      "loss": 3.2062,
+      "step": 2242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.503841525502175e-05,
+      "loss": 2.6683,
+      "step": 2244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.503378691104323e-05,
+      "loss": 2.5465,
+      "step": 2246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.50291585670647e-05,
+      "loss": 2.3814,
+      "step": 2248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.502453022308618e-05,
+      "loss": 1.9564,
+      "step": 2250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5019901879107654e-05,
+      "loss": 0.1662,
+      "step": 2252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.501527353512913e-05,
+      "loss": 2.2642,
+      "step": 2254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5010645191150605e-05,
+      "loss": 1.3017,
+      "step": 2256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5006016847172084e-05,
+      "loss": 1.9808,
+      "step": 2258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5001388503193556e-05,
+      "loss": 0.8798,
+      "step": 2260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4996760159215035e-05,
+      "loss": 1.9536,
+      "step": 2262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4992131815236513e-05,
+      "loss": 6.3355,
+      "step": 2264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4987503471257985e-05,
+      "loss": 1.8066,
+      "step": 2266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4982875127279464e-05,
+      "loss": 2.4471,
+      "step": 2268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4978246783300936e-05,
+      "loss": 0.8584,
+      "step": 2270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4973618439322415e-05,
+      "loss": 1.5624,
+      "step": 2272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.496899009534389e-05,
+      "loss": 6.2858,
+      "step": 2274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4964361751365366e-05,
+      "loss": 1.1941,
+      "step": 2276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.495973340738684e-05,
+      "loss": 2.4854,
+      "step": 2278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.495510506340832e-05,
+      "loss": 4.911,
+      "step": 2280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.495047671942979e-05,
+      "loss": 2.1383,
+      "step": 2282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.494584837545127e-05,
+      "loss": 0.6882,
+      "step": 2284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.494122003147274e-05,
+      "loss": 1.499,
+      "step": 2286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.493659168749422e-05,
+      "loss": 4.5602,
+      "step": 2288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.493196334351569e-05,
+      "loss": 1.0174,
+      "step": 2290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.492733499953717e-05,
+      "loss": 2.5959,
+      "step": 2292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.492270665555864e-05,
+      "loss": 4.0921,
+      "step": 2294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.491807831158012e-05,
+      "loss": 3.4905,
+      "step": 2296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.491344996760159e-05,
+      "loss": 2.3414,
+      "step": 2298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.490882162362307e-05,
+      "loss": 0.6486,
+      "step": 2300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.490419327964454e-05,
+      "loss": 3.1082,
+      "step": 2302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.489956493566602e-05,
+      "loss": 3.1168,
+      "step": 2304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48949365916875e-05,
+      "loss": 4.6532,
+      "step": 2306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.489030824770897e-05,
+      "loss": 1.9955,
+      "step": 2308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.488567990373045e-05,
+      "loss": 5.8696,
+      "step": 2310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4881051559751924e-05,
+      "loss": 4.8305,
+      "step": 2312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48764232157734e-05,
+      "loss": 1.1393,
+      "step": 2314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4871794871794874e-05,
+      "loss": 1.3943,
+      "step": 2316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.486716652781635e-05,
+      "loss": 0.8162,
+      "step": 2318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4862538183837825e-05,
+      "loss": 2.7841,
+      "step": 2320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4857909839859304e-05,
+      "loss": 5.2569,
+      "step": 2322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4853281495880776e-05,
+      "loss": 3.4046,
+      "step": 2324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4848653151902255e-05,
+      "loss": 1.3027,
+      "step": 2326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.484402480792373e-05,
+      "loss": 0.866,
+      "step": 2328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4839396463945206e-05,
+      "loss": 2.8791,
+      "step": 2330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.483476811996668e-05,
+      "loss": 0.1113,
+      "step": 2332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.483013977598816e-05,
+      "loss": 2.381,
+      "step": 2334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.482551143200963e-05,
+      "loss": 3.0403,
+      "step": 2336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.482088308803111e-05,
+      "loss": 1.4951,
+      "step": 2338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.481625474405258e-05,
+      "loss": 2.0979,
+      "step": 2340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.481162640007406e-05,
+      "loss": 2.0601,
+      "step": 2342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.480699805609554e-05,
+      "loss": 0.2744,
+      "step": 2344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.480236971211701e-05,
+      "loss": 7.8202,
+      "step": 2346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.479774136813848e-05,
+      "loss": 1.4182,
+      "step": 2348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.479311302415995e-05,
+      "loss": 5.8921,
+      "step": 2350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.478848468018143e-05,
+      "loss": 3.8961,
+      "step": 2352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4783856336202904e-05,
+      "loss": 2.173,
+      "step": 2354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.477922799222438e-05,
+      "loss": 1.2978,
+      "step": 2356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4774599648245855e-05,
+      "loss": 2.4788,
+      "step": 2358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4769971304267334e-05,
+      "loss": 3.9309,
+      "step": 2360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4765342960288806e-05,
+      "loss": 3.1497,
+      "step": 2362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4760714616310285e-05,
+      "loss": 1.5971,
+      "step": 2364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.475608627233176e-05,
+      "loss": 1.2024,
+      "step": 2366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4751457928353236e-05,
+      "loss": 0.681,
+      "step": 2368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4746829584374714e-05,
+      "loss": 4.8842,
+      "step": 2370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4742201240396186e-05,
+      "loss": 5.752,
+      "step": 2372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4737572896417665e-05,
+      "loss": 6.2372,
+      "step": 2374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.473294455243914e-05,
+      "loss": 4.3843,
+      "step": 2376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4728316208460616e-05,
+      "loss": 4.17,
+      "step": 2378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.472368786448209e-05,
+      "loss": 5.429,
+      "step": 2380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.471905952050357e-05,
+      "loss": 1.4702,
+      "step": 2382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.471443117652504e-05,
+      "loss": 4.9791,
+      "step": 2384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.470980283254652e-05,
+      "loss": 3.6112,
+      "step": 2386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.470517448856799e-05,
+      "loss": 2.094,
+      "step": 2388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.470054614458947e-05,
+      "loss": 3.3698,
+      "step": 2390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.469591780061094e-05,
+      "loss": 1.9098,
+      "step": 2392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.469128945663242e-05,
+      "loss": 2.7917,
+      "step": 2394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.468666111265389e-05,
+      "loss": 0.3426,
+      "step": 2396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.468203276867537e-05,
+      "loss": 3.0689,
+      "step": 2398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.467740442469684e-05,
+      "loss": 1.8702,
+      "step": 2400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.467277608071832e-05,
+      "loss": 0.0791,
+      "step": 2402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.466814773673979e-05,
+      "loss": 1.2237,
+      "step": 2404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.466351939276127e-05,
+      "loss": 1.5561,
+      "step": 2406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4658891048782744e-05,
+      "loss": 2.407,
+      "step": 2408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.465426270480422e-05,
+      "loss": 0.0041,
+      "step": 2410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.46496343608257e-05,
+      "loss": 2.7705,
+      "step": 2412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4645006016847174e-05,
+      "loss": 1.0757,
+      "step": 2414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.464037767286865e-05,
+      "loss": 3.2402,
+      "step": 2416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4635749328890125e-05,
+      "loss": 1.8926,
+      "step": 2418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.46311209849116e-05,
+      "loss": 2.0275,
+      "step": 2420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4626492640933075e-05,
+      "loss": 1.0583,
+      "step": 2422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4621864296954554e-05,
+      "loss": 0.325,
+      "step": 2424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4617235952976026e-05,
+      "loss": 1.194,
+      "step": 2426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4612607608997505e-05,
+      "loss": 1.2236,
+      "step": 2428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.460797926501898e-05,
+      "loss": 2.2921,
+      "step": 2430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4603350921040456e-05,
+      "loss": 0.1005,
+      "step": 2432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.459872257706193e-05,
+      "loss": 0.2327,
+      "step": 2434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.459409423308341e-05,
+      "loss": 5.9419,
+      "step": 2436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.458946588910488e-05,
+      "loss": 0.0061,
+      "step": 2438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.458483754512636e-05,
+      "loss": 0.2648,
+      "step": 2440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.458020920114783e-05,
+      "loss": 3.4569,
+      "step": 2442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.457558085716931e-05,
+      "loss": 6.1861,
+      "step": 2444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.457095251319078e-05,
+      "loss": 5.4836,
+      "step": 2446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.456632416921226e-05,
+      "loss": 1.8945,
+      "step": 2448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.456169582523373e-05,
+      "loss": 1.9144,
+      "step": 2450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.455706748125521e-05,
+      "loss": 3.5867,
+      "step": 2452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.455243913727669e-05,
+      "loss": 3.9415,
+      "step": 2454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.454781079329816e-05,
+      "loss": 3.6372,
+      "step": 2456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.454318244931964e-05,
+      "loss": 1.1826,
+      "step": 2458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.453855410534111e-05,
+      "loss": 1.5229,
+      "step": 2460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.453392576136259e-05,
+      "loss": 2.71,
+      "step": 2462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.452929741738406e-05,
+      "loss": 1.2933,
+      "step": 2464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.452466907340554e-05,
+      "loss": 3.0796,
+      "step": 2466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4520040729427014e-05,
+      "loss": 4.9143,
+      "step": 2468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.451541238544849e-05,
+      "loss": 0.6296,
+      "step": 2470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4510784041469964e-05,
+      "loss": 4.4411,
+      "step": 2472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.450615569749144e-05,
+      "loss": 0.9448,
+      "step": 2474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4501527353512915e-05,
+      "loss": 6.1768,
+      "step": 2476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4496899009534394e-05,
+      "loss": 0.514,
+      "step": 2478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4492270665555866e-05,
+      "loss": 0.9086,
+      "step": 2480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4487642321577345e-05,
+      "loss": 5.5642,
+      "step": 2482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.448301397759882e-05,
+      "loss": 1.4092,
+      "step": 2484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4478385633620296e-05,
+      "loss": 1.4034,
+      "step": 2486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.447375728964177e-05,
+      "loss": 7.0888,
+      "step": 2488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4469128945663247e-05,
+      "loss": 5.1851,
+      "step": 2490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4464500601684725e-05,
+      "loss": 4.5908,
+      "step": 2492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.445987225770619e-05,
+      "loss": 1.0254,
+      "step": 2494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.445524391372767e-05,
+      "loss": 3.7667,
+      "step": 2496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.445061556974914e-05,
+      "loss": 2.3717,
+      "step": 2498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.444598722577062e-05,
+      "loss": 0.2581,
+      "step": 2500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.444135888179209e-05,
+      "loss": 1.4121,
+      "step": 2502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.443673053781357e-05,
+      "loss": 6.1679,
+      "step": 2504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.443210219383504e-05,
+      "loss": 2.0511,
+      "step": 2506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.442747384985652e-05,
+      "loss": 1.4458,
+      "step": 2508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4422845505877994e-05,
+      "loss": 1.7507,
+      "step": 2510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.441821716189947e-05,
+      "loss": 1.6248,
+      "step": 2512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4413588817920945e-05,
+      "loss": 0.303,
+      "step": 2514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4408960473942424e-05,
+      "loss": 2.1568,
+      "step": 2516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.44043321299639e-05,
+      "loss": 2.6031,
+      "step": 2518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4399703785985375e-05,
+      "loss": 0.0373,
+      "step": 2520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.439507544200685e-05,
+      "loss": 0.6339,
+      "step": 2522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4390447098028325e-05,
+      "loss": 1.3996,
+      "step": 2524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4385818754049804e-05,
+      "loss": 2.5112,
+      "step": 2526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4381190410071276e-05,
+      "loss": 5.8007,
+      "step": 2528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4376562066092755e-05,
+      "loss": 5.9392,
+      "step": 2530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.437193372211423e-05,
+      "loss": 1.1424,
+      "step": 2532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4367305378135706e-05,
+      "loss": 5.5354,
+      "step": 2534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.436267703415718e-05,
+      "loss": 0.9002,
+      "step": 2536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.435804869017866e-05,
+      "loss": 1.965,
+      "step": 2538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.435342034620013e-05,
+      "loss": 1.4898,
+      "step": 2540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.434879200222161e-05,
+      "loss": 6.6592,
+      "step": 2542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.434416365824308e-05,
+      "loss": 0.9809,
+      "step": 2544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.433953531426456e-05,
+      "loss": 2.5261,
+      "step": 2546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.433490697028603e-05,
+      "loss": 2.4459,
+      "step": 2548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.433027862630751e-05,
+      "loss": 1.0395,
+      "step": 2550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.432565028232898e-05,
+      "loss": 0.3818,
+      "step": 2552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.432102193835046e-05,
+      "loss": 3.5728,
+      "step": 2554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.431639359437193e-05,
+      "loss": 3.5446,
+      "step": 2556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.431176525039341e-05,
+      "loss": 3.4712,
+      "step": 2558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.430713690641489e-05,
+      "loss": 0.6659,
+      "step": 2560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.430250856243636e-05,
+      "loss": 4.5963,
+      "step": 2562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.429788021845784e-05,
+      "loss": 0.9314,
+      "step": 2564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.429325187447931e-05,
+      "loss": 2.7721,
+      "step": 2566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.428862353050079e-05,
+      "loss": 0.2912,
+      "step": 2568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4283995186522264e-05,
+      "loss": 3.0583,
+      "step": 2570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.427936684254374e-05,
+      "loss": 2.545,
+      "step": 2572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4274738498565214e-05,
+      "loss": 3.8111,
+      "step": 2574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.427011015458669e-05,
+      "loss": 3.7708,
+      "step": 2576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4265481810608165e-05,
+      "loss": 5.1702,
+      "step": 2578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4260853466629644e-05,
+      "loss": 0.6992,
+      "step": 2580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4256225122651116e-05,
+      "loss": 2.4681,
+      "step": 2582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4251596778672595e-05,
+      "loss": 4.0294,
+      "step": 2584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.424696843469407e-05,
+      "loss": 2.8139,
+      "step": 2586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4242340090715546e-05,
+      "loss": 1.4004,
+      "step": 2588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.423771174673702e-05,
+      "loss": 3.3789,
+      "step": 2590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4233083402758497e-05,
+      "loss": 2.332,
+      "step": 2592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.422845505877997e-05,
+      "loss": 1.0721,
+      "step": 2594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.422382671480145e-05,
+      "loss": 0.4702,
+      "step": 2596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4219198370822926e-05,
+      "loss": 3.6235,
+      "step": 2598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42145700268444e-05,
+      "loss": 2.7316,
+      "step": 2600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.420994168286588e-05,
+      "loss": 0.0648,
+      "step": 2602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.420531333888735e-05,
+      "loss": 3.0792,
+      "step": 2604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.420068499490883e-05,
+      "loss": 0.207,
+      "step": 2606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.41960566509303e-05,
+      "loss": 0.0044,
+      "step": 2608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.419142830695178e-05,
+      "loss": 7.9926,
+      "step": 2610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.418679996297325e-05,
+      "loss": 3.725,
+      "step": 2612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.418217161899473e-05,
+      "loss": 0.0093,
+      "step": 2614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.41775432750162e-05,
+      "loss": 0.4651,
+      "step": 2616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.417291493103768e-05,
+      "loss": 2.9808,
+      "step": 2618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.416828658705915e-05,
+      "loss": 0.3609,
+      "step": 2620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.416365824308063e-05,
+      "loss": 0.5582,
+      "step": 2622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4159029899102103e-05,
+      "loss": 0.2891,
+      "step": 2624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.415440155512358e-05,
+      "loss": 3.892,
+      "step": 2626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4149773211145054e-05,
+      "loss": 0.775,
+      "step": 2628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.414514486716653e-05,
+      "loss": 2.5052,
+      "step": 2630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4140516523188005e-05,
+      "loss": 3.1075,
+      "step": 2632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4135888179209484e-05,
+      "loss": 4.7003,
+      "step": 2634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4131259835230956e-05,
+      "loss": 2.4242,
+      "step": 2636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.412663149125243e-05,
+      "loss": 7.5006,
+      "step": 2638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.412200314727391e-05,
+      "loss": 4.1007,
+      "step": 2640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.411737480329538e-05,
+      "loss": 1.8945,
+      "step": 2642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.411274645931686e-05,
+      "loss": 2.61,
+      "step": 2644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.410811811533833e-05,
+      "loss": 4.0779,
+      "step": 2646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.410348977135981e-05,
+      "loss": 1.3684,
+      "step": 2648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.409886142738128e-05,
+      "loss": 2.0648,
+      "step": 2650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.409423308340276e-05,
+      "loss": 5.6303,
+      "step": 2652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.408960473942423e-05,
+      "loss": 4.1201,
+      "step": 2654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.408497639544571e-05,
+      "loss": 2.9721,
+      "step": 2656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.408034805146718e-05,
+      "loss": 3.1875,
+      "step": 2658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.407571970748866e-05,
+      "loss": 3.6593,
+      "step": 2660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.407109136351013e-05,
+      "loss": 2.4537,
+      "step": 2662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.406646301953161e-05,
+      "loss": 3.2957,
+      "step": 2664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.406183467555309e-05,
+      "loss": 3.8032,
+      "step": 2666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.405720633157456e-05,
+      "loss": 2.9254,
+      "step": 2668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.405257798759604e-05,
+      "loss": 2.7486,
+      "step": 2670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4047949643617514e-05,
+      "loss": 1.1837,
+      "step": 2672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.404332129963899e-05,
+      "loss": 2.7018,
+      "step": 2674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4038692955660464e-05,
+      "loss": 3.6432,
+      "step": 2676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.403406461168194e-05,
+      "loss": 3.5784,
+      "step": 2678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4029436267703415e-05,
+      "loss": 1.4645,
+      "step": 2680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4024807923724894e-05,
+      "loss": 2.8162,
+      "step": 2682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4020179579746366e-05,
+      "loss": 5.4228,
+      "step": 2684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4015551235767845e-05,
+      "loss": 4.3235,
+      "step": 2686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.401092289178932e-05,
+      "loss": 3.1405,
+      "step": 2688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4006294547810796e-05,
+      "loss": 3.6612,
+      "step": 2690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.400166620383227e-05,
+      "loss": 1.943,
+      "step": 2692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.399703785985375e-05,
+      "loss": 2.3171,
+      "step": 2694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.399240951587522e-05,
+      "loss": 2.9479,
+      "step": 2696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.39877811718967e-05,
+      "loss": 5.2004,
+      "step": 2698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.398315282791817e-05,
+      "loss": 2.3372,
+      "step": 2700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.397852448393965e-05,
+      "loss": 1.9641,
+      "step": 2702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.397389613996113e-05,
+      "loss": 5.2737,
+      "step": 2704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.39692677959826e-05,
+      "loss": 0.7862,
+      "step": 2706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.396463945200408e-05,
+      "loss": 3.2986,
+      "step": 2708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.396001110802555e-05,
+      "loss": 4.3655,
+      "step": 2710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.395538276404703e-05,
+      "loss": 1.5156,
+      "step": 2712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.39507544200685e-05,
+      "loss": 3.8107,
+      "step": 2714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.394612607608998e-05,
+      "loss": 3.4354,
+      "step": 2716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.394149773211145e-05,
+      "loss": 2.8509,
+      "step": 2718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.393686938813293e-05,
+      "loss": 4.3572,
+      "step": 2720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.39322410441544e-05,
+      "loss": 0.0784,
+      "step": 2722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.392761270017588e-05,
+      "loss": 1.6573,
+      "step": 2724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3922984356197353e-05,
+      "loss": 0.7686,
+      "step": 2726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.391835601221883e-05,
+      "loss": 6.2967,
+      "step": 2728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3913727668240304e-05,
+      "loss": 4.4603,
+      "step": 2730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.390909932426178e-05,
+      "loss": 1.0802,
+      "step": 2732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3904470980283255e-05,
+      "loss": 4.3052,
+      "step": 2734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3899842636304734e-05,
+      "loss": 2.7789,
+      "step": 2736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3895214292326206e-05,
+      "loss": 0.0174,
+      "step": 2738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3890585948347685e-05,
+      "loss": 0.5841,
+      "step": 2740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.388595760436916e-05,
+      "loss": 3.9594,
+      "step": 2742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3881329260390636e-05,
+      "loss": 3.4283,
+      "step": 2744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3876700916412114e-05,
+      "loss": 0.0334,
+      "step": 2746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3872072572433587e-05,
+      "loss": 3.6736,
+      "step": 2748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3867444228455065e-05,
+      "loss": 3.993,
+      "step": 2750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.386281588447654e-05,
+      "loss": 2.7755,
+      "step": 2752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3858187540498016e-05,
+      "loss": 2.6507,
+      "step": 2754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.385355919651949e-05,
+      "loss": 2.7294,
+      "step": 2756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.384893085254097e-05,
+      "loss": 1.3126,
+      "step": 2758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.384430250856244e-05,
+      "loss": 2.0301,
+      "step": 2760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.383967416458392e-05,
+      "loss": 2.633,
+      "step": 2762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.383504582060539e-05,
+      "loss": 0.5288,
+      "step": 2764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.383041747662687e-05,
+      "loss": 1.1458,
+      "step": 2766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.382578913264834e-05,
+      "loss": 2.0533,
+      "step": 2768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.382116078866982e-05,
+      "loss": 1.4202,
+      "step": 2770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.381653244469129e-05,
+      "loss": 5.0797,
+      "step": 2772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.381190410071277e-05,
+      "loss": 1.7898,
+      "step": 2774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.380727575673424e-05,
+      "loss": 1.5671,
+      "step": 2776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.380264741275572e-05,
+      "loss": 2.1318,
+      "step": 2778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.379801906877719e-05,
+      "loss": 1.4095,
+      "step": 2780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.379339072479867e-05,
+      "loss": 0.0043,
+      "step": 2782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3788762380820144e-05,
+      "loss": 1.1215,
+      "step": 2784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3784134036841616e-05,
+      "loss": 1.2865,
+      "step": 2786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3779505692863095e-05,
+      "loss": 5.3486,
+      "step": 2788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.377487734888457e-05,
+      "loss": 1.8235,
+      "step": 2790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3770249004906046e-05,
+      "loss": 1.3124,
+      "step": 2792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.376562066092752e-05,
+      "loss": 4.5353,
+      "step": 2794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3760992316949e-05,
+      "loss": 0.7145,
+      "step": 2796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.375636397297047e-05,
+      "loss": 3.7446,
+      "step": 2798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.375173562899195e-05,
+      "loss": 0.9431,
+      "step": 2800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.374710728501342e-05,
+      "loss": 1.3472,
+      "step": 2802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.37424789410349e-05,
+      "loss": 0.8382,
+      "step": 2804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.373785059705637e-05,
+      "loss": 3.288,
+      "step": 2806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.373322225307785e-05,
+      "loss": 2.5933,
+      "step": 2808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.372859390909932e-05,
+      "loss": 1.5494,
+      "step": 2810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.37239655651208e-05,
+      "loss": 0.9717,
+      "step": 2812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.371933722114228e-05,
+      "loss": 2.3439,
+      "step": 2814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.371470887716375e-05,
+      "loss": 2.4958,
+      "step": 2816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.371008053318523e-05,
+      "loss": 4.3141,
+      "step": 2818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.37054521892067e-05,
+      "loss": 2.8412,
+      "step": 2820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.370082384522818e-05,
+      "loss": 1.4193,
+      "step": 2822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.369619550124965e-05,
+      "loss": 3.127,
+      "step": 2824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.369156715727113e-05,
+      "loss": 3.2877,
+      "step": 2826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3686938813292603e-05,
+      "loss": 0.975,
+      "step": 2828
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.368231046931408e-05,
+      "loss": 1.605,
+      "step": 2830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3677682125335554e-05,
+      "loss": 0.5697,
+      "step": 2832
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.367305378135703e-05,
+      "loss": 4.5788,
+      "step": 2834
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3668425437378505e-05,
+      "loss": 0.7533,
+      "step": 2836
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3663797093399984e-05,
+      "loss": 2.2919,
+      "step": 2838
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3659168749421456e-05,
+      "loss": 0.5652,
+      "step": 2840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3654540405442935e-05,
+      "loss": 6.8378,
+      "step": 2842
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.364991206146441e-05,
+      "loss": 1.3532,
+      "step": 2844
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3645283717485886e-05,
+      "loss": 2.8194,
+      "step": 2846
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.364065537350736e-05,
+      "loss": 5.205,
+      "step": 2848
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3636027029528837e-05,
+      "loss": 3.8505,
+      "step": 2850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3631398685550315e-05,
+      "loss": 2.6481,
+      "step": 2852
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.362677034157179e-05,
+      "loss": 2.8456,
+      "step": 2854
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3622141997593266e-05,
+      "loss": 4.1129,
+      "step": 2856
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.361751365361474e-05,
+      "loss": 4.2937,
+      "step": 2858
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.361288530963622e-05,
+      "loss": 2.4405,
+      "step": 2860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360825696565769e-05,
+      "loss": 1.2435,
+      "step": 2862
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360362862167917e-05,
+      "loss": 2.9837,
+      "step": 2864
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.359900027770064e-05,
+      "loss": 1.0911,
+      "step": 2866
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.359437193372212e-05,
+      "loss": 1.2255,
+      "step": 2868
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 0.8968,
+      "step": 2870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.358511524576507e-05,
+      "loss": 4.6966,
+      "step": 2872
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.358048690178654e-05,
+      "loss": 2.5355,
+      "step": 2874
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.357585855780802e-05,
+      "loss": 2.2348,
+      "step": 2876
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.357123021382949e-05,
+      "loss": 3.9803,
+      "step": 2878
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.356660186985097e-05,
+      "loss": 1.894,
+      "step": 2880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.356197352587244e-05,
+      "loss": 1.7825,
+      "step": 2882
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.355734518189392e-05,
+      "loss": 0.8478,
+      "step": 2884
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3552716837915394e-05,
+      "loss": 7.4051,
+      "step": 2886
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.354808849393687e-05,
+      "loss": 3.6924,
+      "step": 2888
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3543460149958345e-05,
+      "loss": 2.776,
+      "step": 2890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3538831805979824e-05,
+      "loss": 2.752,
+      "step": 2892
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.35342034620013e-05,
+      "loss": 1.3285,
+      "step": 2894
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3529575118022775e-05,
+      "loss": 1.9698,
+      "step": 2896
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3524946774044254e-05,
+      "loss": 1.5089,
+      "step": 2898
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3520318430065726e-05,
+      "loss": 1.1621,
+      "step": 2900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3515690086087204e-05,
+      "loss": 3.0624,
+      "step": 2902
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3511061742108676e-05,
+      "loss": 3.4497,
+      "step": 2904
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3506433398130155e-05,
+      "loss": 1.9959,
+      "step": 2906
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.350180505415163e-05,
+      "loss": 6.8955,
+      "step": 2908
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3497176710173106e-05,
+      "loss": 2.1801,
+      "step": 2910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.349254836619458e-05,
+      "loss": 2.7468,
+      "step": 2912
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.348792002221606e-05,
+      "loss": 2.1356,
+      "step": 2914
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.348329167823753e-05,
+      "loss": 2.996,
+      "step": 2916
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.347866333425901e-05,
+      "loss": 1.6563,
+      "step": 2918
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.347403499028048e-05,
+      "loss": 2.8704,
+      "step": 2920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.346940664630196e-05,
+      "loss": 1.5192,
+      "step": 2922
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.346477830232343e-05,
+      "loss": 1.6347,
+      "step": 2924
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.346014995834491e-05,
+      "loss": 3.7213,
+      "step": 2926
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.345552161436638e-05,
+      "loss": 2.5478,
+      "step": 2928
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3450893270387854e-05,
+      "loss": 1.9151,
+      "step": 2930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.344626492640933e-05,
+      "loss": 3.5758,
+      "step": 2932
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3441636582430804e-05,
+      "loss": 1.8049,
+      "step": 2934
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.343700823845228e-05,
+      "loss": 3.7361,
+      "step": 2936
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3432379894473755e-05,
+      "loss": 1.9746,
+      "step": 2938
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3427751550495234e-05,
+      "loss": 1.2504,
+      "step": 2940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3423123206516706e-05,
+      "loss": 2.9904,
+      "step": 2942
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3418494862538185e-05,
+      "loss": 4.0428,
+      "step": 2944
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.341386651855966e-05,
+      "loss": 3.0578,
+      "step": 2946
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3409238174581136e-05,
+      "loss": 3.4413,
+      "step": 2948
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.340460983060261e-05,
+      "loss": 2.0494,
+      "step": 2950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3399981486624087e-05,
+      "loss": 1.621,
+      "step": 2952
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.339535314264556e-05,
+      "loss": 2.9422,
+      "step": 2954
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.339072479866704e-05,
+      "loss": 0.8094,
+      "step": 2956
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3386096454688516e-05,
+      "loss": 1.1235,
+      "step": 2958
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.338146811070999e-05,
+      "loss": 0.8384,
+      "step": 2960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.337683976673147e-05,
+      "loss": 1.6409,
+      "step": 2962
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.337221142275294e-05,
+      "loss": 1.2324,
+      "step": 2964
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.336758307877442e-05,
+      "loss": 3.5691,
+      "step": 2966
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.336295473479589e-05,
+      "loss": 5.449,
+      "step": 2968
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.335832639081737e-05,
+      "loss": 1.2248,
+      "step": 2970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.335369804683884e-05,
+      "loss": 2.1359,
+      "step": 2972
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.334906970286032e-05,
+      "loss": 2.8838,
+      "step": 2974
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.334444135888179e-05,
+      "loss": 4.8468,
+      "step": 2976
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.333981301490327e-05,
+      "loss": 3.0735,
+      "step": 2978
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.333518467092474e-05,
+      "loss": 2.0088,
+      "step": 2980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.333055632694622e-05,
+      "loss": 6.8113,
+      "step": 2982
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.332592798296769e-05,
+      "loss": 1.7866,
+      "step": 2984
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.332129963898917e-05,
+      "loss": 1.5066,
+      "step": 2986
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3316671295010644e-05,
+      "loss": 3.4974,
+      "step": 2988
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.331204295103212e-05,
+      "loss": 2.2421,
+      "step": 2990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3307414607053595e-05,
+      "loss": 2.6757,
+      "step": 2992
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3302786263075074e-05,
+      "loss": 2.9015,
+      "step": 2994
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3298157919096546e-05,
+      "loss": 1.4698,
+      "step": 2996
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3293529575118025e-05,
+      "loss": 2.3329,
+      "step": 2998
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3288901231139504e-05,
+      "loss": 1.6533,
+      "step": 3000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3284272887160976e-05,
+      "loss": 0.5187,
+      "step": 3002
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3279644543182454e-05,
+      "loss": 4.2235,
+      "step": 3004
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3275016199203926e-05,
+      "loss": 1.8987,
+      "step": 3006
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3270387855225405e-05,
+      "loss": 1.5289,
+      "step": 3008
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.326575951124688e-05,
+      "loss": 0.0567,
+      "step": 3010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3261131167268356e-05,
+      "loss": 9.8389,
+      "step": 3012
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.325650282328983e-05,
+      "loss": 2.1317,
+      "step": 3014
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.325187447931131e-05,
+      "loss": 2.8164,
+      "step": 3016
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.324724613533278e-05,
+      "loss": 1.89,
+      "step": 3018
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.324261779135426e-05,
+      "loss": 3.6251,
+      "step": 3020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.323798944737573e-05,
+      "loss": 2.2447,
+      "step": 3022
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.323336110339721e-05,
+      "loss": 6.4708,
+      "step": 3024
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.322873275941868e-05,
+      "loss": 2.7006,
+      "step": 3026
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.322410441544016e-05,
+      "loss": 1.7226,
+      "step": 3028
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.321947607146163e-05,
+      "loss": 0.0546,
+      "step": 3030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.321484772748311e-05,
+      "loss": 1.5409,
+      "step": 3032
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.321021938350458e-05,
+      "loss": 7.4564,
+      "step": 3034
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.320559103952606e-05,
+      "loss": 0.9203,
+      "step": 3036
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.320096269554753e-05,
+      "loss": 0.6046,
+      "step": 3038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.319633435156901e-05,
+      "loss": 1.8553,
+      "step": 3040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.319170600759049e-05,
+      "loss": 0.7454,
+      "step": 3042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.318707766361196e-05,
+      "loss": 7.2183,
+      "step": 3044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.318244931963344e-05,
+      "loss": 3.8499,
+      "step": 3046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3177820975654914e-05,
+      "loss": 1.0101,
+      "step": 3048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.317319263167639e-05,
+      "loss": 0.6399,
+      "step": 3050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3168564287697865e-05,
+      "loss": 2.9095,
+      "step": 3052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3163935943719343e-05,
+      "loss": 6.8418,
+      "step": 3054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3159307599740815e-05,
+      "loss": 3.5289,
+      "step": 3056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3154679255762294e-05,
+      "loss": 2.098,
+      "step": 3058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3150050911783766e-05,
+      "loss": 2.8032,
+      "step": 3060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3145422567805245e-05,
+      "loss": 2.3119,
+      "step": 3062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.314079422382672e-05,
+      "loss": 1.8421,
+      "step": 3064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3136165879848196e-05,
+      "loss": 3.1715,
+      "step": 3066
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.313153753586967e-05,
+      "loss": 3.8869,
+      "step": 3068
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.312690919189115e-05,
+      "loss": 3.4876,
+      "step": 3070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.312228084791262e-05,
+      "loss": 0.4605,
+      "step": 3072
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.311765250393409e-05,
+      "loss": 4.0718,
+      "step": 3074
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.311302415995557e-05,
+      "loss": 1.8317,
+      "step": 3076
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.310839581597704e-05,
+      "loss": 0.2,
+      "step": 3078
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.310376747199852e-05,
+      "loss": 1.9115,
+      "step": 3080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.309913912801999e-05,
+      "loss": 2.8866,
+      "step": 3082
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.309451078404147e-05,
+      "loss": 0.0458,
+      "step": 3084
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3089882440062943e-05,
+      "loss": 5.833,
+      "step": 3086
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.308525409608442e-05,
+      "loss": 1.1995,
+      "step": 3088
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3080625752105894e-05,
+      "loss": 0.9039,
+      "step": 3090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.307599740812737e-05,
+      "loss": 5.8866,
+      "step": 3092
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3071369064148845e-05,
+      "loss": 0.7266,
+      "step": 3094
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3066740720170324e-05,
+      "loss": 0.6991,
+      "step": 3096
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3062112376191796e-05,
+      "loss": 1.9344,
+      "step": 3098
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3057484032213275e-05,
+      "loss": 1.4056,
+      "step": 3100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.305285568823475e-05,
+      "loss": 1.3197,
+      "step": 3102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3048227344256226e-05,
+      "loss": 2.046,
+      "step": 3104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3043599000277704e-05,
+      "loss": 0.0402,
+      "step": 3106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3038970656299176e-05,
+      "loss": 0.374,
+      "step": 3108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3034342312320655e-05,
+      "loss": 3.9081,
+      "step": 3110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.302971396834213e-05,
+      "loss": 7.0488,
+      "step": 3112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3025085624363606e-05,
+      "loss": 7.4802,
+      "step": 3114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.302045728038508e-05,
+      "loss": 2.5635,
+      "step": 3116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.301582893640656e-05,
+      "loss": 0.0342,
+      "step": 3118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.301120059242803e-05,
+      "loss": 1.1016,
+      "step": 3120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.300657224844951e-05,
+      "loss": 1.1263,
+      "step": 3122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.300194390447098e-05,
+      "loss": 4.8935,
+      "step": 3124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.299731556049246e-05,
+      "loss": 2.4525,
+      "step": 3126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.299268721651393e-05,
+      "loss": 2.3383,
+      "step": 3128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.298805887253541e-05,
+      "loss": 1.8077,
+      "step": 3130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.298343052855688e-05,
+      "loss": 0.4661,
+      "step": 3132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.297880218457836e-05,
+      "loss": 7.0783,
+      "step": 3134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.297417384059983e-05,
+      "loss": 3.5726,
+      "step": 3136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.296954549662131e-05,
+      "loss": 1.1502,
+      "step": 3138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.296491715264278e-05,
+      "loss": 2.4616,
+      "step": 3140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.296028880866426e-05,
+      "loss": 4.9733,
+      "step": 3142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2955660464685734e-05,
+      "loss": 2.754,
+      "step": 3144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.295103212070721e-05,
+      "loss": 3.1002,
+      "step": 3146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.294640377672869e-05,
+      "loss": 6.7086,
+      "step": 3148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2941775432750164e-05,
+      "loss": 3.1922,
+      "step": 3150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.293714708877164e-05,
+      "loss": 4.0187,
+      "step": 3152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2932518744793115e-05,
+      "loss": 2.4182,
+      "step": 3154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2927890400814593e-05,
+      "loss": 3.0794,
+      "step": 3156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2923262056836065e-05,
+      "loss": 1.2806,
+      "step": 3158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2918633712857544e-05,
+      "loss": 1.2764,
+      "step": 3160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2914005368879016e-05,
+      "loss": 4.7056,
+      "step": 3162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2909377024900495e-05,
+      "loss": 2.0243,
+      "step": 3164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.290474868092197e-05,
+      "loss": 4.0551,
+      "step": 3166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2900120336943446e-05,
+      "loss": 0.6186,
+      "step": 3168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.289549199296492e-05,
+      "loss": 2.2275,
+      "step": 3170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.28908636489864e-05,
+      "loss": 2.5112,
+      "step": 3172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.288623530500787e-05,
+      "loss": 1.2749,
+      "step": 3174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.288160696102935e-05,
+      "loss": 3.9905,
+      "step": 3176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.287697861705082e-05,
+      "loss": 3.2244,
+      "step": 3178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.28723502730723e-05,
+      "loss": 6.6179,
+      "step": 3180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.286772192909377e-05,
+      "loss": 1.2725,
+      "step": 3182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.286309358511525e-05,
+      "loss": 0.0003,
+      "step": 3184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.285846524113673e-05,
+      "loss": 1.1506,
+      "step": 3186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.28538368971582e-05,
+      "loss": 2.4248,
+      "step": 3188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.284920855317968e-05,
+      "loss": 1.6929,
+      "step": 3190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.284458020920115e-05,
+      "loss": 2.8815,
+      "step": 3192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.283995186522263e-05,
+      "loss": 4.4427,
+      "step": 3194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.28353235212441e-05,
+      "loss": 3.1305,
+      "step": 3196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.283069517726558e-05,
+      "loss": 0.8681,
+      "step": 3198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.282606683328705e-05,
+      "loss": 1.4018,
+      "step": 3200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.282143848930853e-05,
+      "loss": 1.5007,
+      "step": 3202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2816810145330004e-05,
+      "loss": 4.7174,
+      "step": 3204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.281218180135148e-05,
+      "loss": 3.5461,
+      "step": 3206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2807553457372954e-05,
+      "loss": 3.2717,
+      "step": 3208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.280292511339443e-05,
+      "loss": 0.0314,
+      "step": 3210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2798296769415905e-05,
+      "loss": 0.5393,
+      "step": 3212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2793668425437384e-05,
+      "loss": 0.0365,
+      "step": 3214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2789040081458856e-05,
+      "loss": 2.0317,
+      "step": 3216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.278441173748033e-05,
+      "loss": 0.8583,
+      "step": 3218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.277978339350181e-05,
+      "loss": 6.0489,
+      "step": 3220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.277515504952328e-05,
+      "loss": 2.9371,
+      "step": 3222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.277052670554476e-05,
+      "loss": 1.0969,
+      "step": 3224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.276589836156623e-05,
+      "loss": 0.9496,
+      "step": 3226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.276127001758771e-05,
+      "loss": 2.5164,
+      "step": 3228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.275664167360918e-05,
+      "loss": 8.0292,
+      "step": 3230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.275201332963066e-05,
+      "loss": 2.4253,
+      "step": 3232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.274738498565213e-05,
+      "loss": 3.5435,
+      "step": 3234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.274275664167361e-05,
+      "loss": 4.0552,
+      "step": 3236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.273812829769508e-05,
+      "loss": 3.2006,
+      "step": 3238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.273349995371656e-05,
+      "loss": 3.2304,
+      "step": 3240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.272887160973803e-05,
+      "loss": 1.9212,
+      "step": 3242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.272424326575951e-05,
+      "loss": 2.2574,
+      "step": 3244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2719614921780984e-05,
+      "loss": 5.3556,
+      "step": 3246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.271498657780246e-05,
+      "loss": 3.1678,
+      "step": 3248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2710358233823935e-05,
+      "loss": 2.3383,
+      "step": 3250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2705729889845414e-05,
+      "loss": 2.4842,
+      "step": 3252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.270110154586689e-05,
+      "loss": 0.0118,
+      "step": 3254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2696473201888365e-05,
+      "loss": 4.3121,
+      "step": 3256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2691844857909843e-05,
+      "loss": 1.8563,
+      "step": 3258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2687216513931316e-05,
+      "loss": 0.1075,
+      "step": 3260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2682588169952794e-05,
+      "loss": 1.7577,
+      "step": 3262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2677959825974266e-05,
+      "loss": 2.1716,
+      "step": 3264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2673331481995745e-05,
+      "loss": 3.6116,
+      "step": 3266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.266870313801722e-05,
+      "loss": 0.2518,
+      "step": 3268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2664074794038696e-05,
+      "loss": 0.7862,
+      "step": 3270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.265944645006017e-05,
+      "loss": 0.2902,
+      "step": 3272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.265481810608165e-05,
+      "loss": 1.3783,
+      "step": 3274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.265018976210312e-05,
+      "loss": 4.187,
+      "step": 3276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26455614181246e-05,
+      "loss": 2.8048,
+      "step": 3278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.264093307414607e-05,
+      "loss": 1.9152,
+      "step": 3280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.263630473016755e-05,
+      "loss": 1.3466,
+      "step": 3282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.263167638618902e-05,
+      "loss": 6.0907,
+      "step": 3284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26270480422105e-05,
+      "loss": 1.5592,
+      "step": 3286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.262241969823197e-05,
+      "loss": 0.7997,
+      "step": 3288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.261779135425345e-05,
+      "loss": 5.2901,
+      "step": 3290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.261316301027493e-05,
+      "loss": 1.6352,
+      "step": 3292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26085346662964e-05,
+      "loss": 3.447,
+      "step": 3294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.260390632231788e-05,
+      "loss": 2.47,
+      "step": 3296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.259927797833935e-05,
+      "loss": 1.4011,
+      "step": 3298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.259464963436083e-05,
+      "loss": 0.186,
+      "step": 3300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.25900212903823e-05,
+      "loss": 2.8851,
+      "step": 3302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.258539294640378e-05,
+      "loss": 0.9522,
+      "step": 3304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2580764602425254e-05,
+      "loss": 1.1656,
+      "step": 3306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.257613625844673e-05,
+      "loss": 6.3406,
+      "step": 3308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2571507914468205e-05,
+      "loss": 3.9567,
+      "step": 3310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.256687957048968e-05,
+      "loss": 1.6652,
+      "step": 3312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2562251226511155e-05,
+      "loss": 1.4155,
+      "step": 3314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2557622882532634e-05,
+      "loss": 4.0032,
+      "step": 3316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2552994538554106e-05,
+      "loss": 1.365,
+      "step": 3318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2548366194575585e-05,
+      "loss": 2.2116,
+      "step": 3320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.254373785059706e-05,
+      "loss": 3.6597,
+      "step": 3322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2539109506618536e-05,
+      "loss": 2.9412,
+      "step": 3324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.253448116264001e-05,
+      "loss": 4.1019,
+      "step": 3326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.252985281866149e-05,
+      "loss": 3.5006,
+      "step": 3328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.252522447468296e-05,
+      "loss": 0.6264,
+      "step": 3330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.252059613070444e-05,
+      "loss": 1.0123,
+      "step": 3332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2515967786725916e-05,
+      "loss": 2.6124,
+      "step": 3334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.251133944274739e-05,
+      "loss": 2.4115,
+      "step": 3336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.250671109876887e-05,
+      "loss": 2.7729,
+      "step": 3338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.250208275479034e-05,
+      "loss": 1.8189,
+      "step": 3340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.249745441081182e-05,
+      "loss": 2.9102,
+      "step": 3342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.249282606683329e-05,
+      "loss": 0.4764,
+      "step": 3344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.248819772285477e-05,
+      "loss": 5.0997,
+      "step": 3346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.248356937887624e-05,
+      "loss": 0.793,
+      "step": 3348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.247894103489772e-05,
+      "loss": 0.7547,
+      "step": 3350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.247431269091919e-05,
+      "loss": 0.959,
+      "step": 3352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.246968434694067e-05,
+      "loss": 1.7352,
+      "step": 3354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.246505600296214e-05,
+      "loss": 0.0378,
+      "step": 3356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.246042765898362e-05,
+      "loss": 0.1231,
+      "step": 3358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2455799315005094e-05,
+      "loss": 1.2512,
+      "step": 3360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.245117097102657e-05,
+      "loss": 2.6686,
+      "step": 3362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2446542627048044e-05,
+      "loss": 4.1809,
+      "step": 3364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2441914283069516e-05,
+      "loss": 1.5655,
+      "step": 3366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2437285939090995e-05,
+      "loss": 5.4001,
+      "step": 3368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.243265759511247e-05,
+      "loss": 3.8123,
+      "step": 3370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2428029251133946e-05,
+      "loss": 1.4871,
+      "step": 3372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.242340090715542e-05,
+      "loss": 0.3414,
+      "step": 3374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.24187725631769e-05,
+      "loss": 2.6262,
+      "step": 3376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.241414421919837e-05,
+      "loss": 0.0363,
+      "step": 3378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.240951587521985e-05,
+      "loss": 0.1372,
+      "step": 3380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.240488753124132e-05,
+      "loss": 0.8853,
+      "step": 3382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.24002591872628e-05,
+      "loss": 0.0597,
+      "step": 3384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.239563084328427e-05,
+      "loss": 1.7297,
+      "step": 3386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.239100249930575e-05,
+      "loss": 4.9774,
+      "step": 3388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.238637415532722e-05,
+      "loss": 1.7487,
+      "step": 3390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.23817458113487e-05,
+      "loss": 0.005,
+      "step": 3392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.237711746737017e-05,
+      "loss": 6.1751,
+      "step": 3394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.237248912339165e-05,
+      "loss": 4.0365,
+      "step": 3396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.236786077941312e-05,
+      "loss": 2.4866,
+      "step": 3398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.23632324354346e-05,
+      "loss": 3.9682,
+      "step": 3400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.235860409145608e-05,
+      "loss": 0.2623,
+      "step": 3402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.235397574747755e-05,
+      "loss": 2.031,
+      "step": 3404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.234934740349903e-05,
+      "loss": 0.2057,
+      "step": 3406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2344719059520504e-05,
+      "loss": 1.7295,
+      "step": 3408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.234009071554198e-05,
+      "loss": 4.2343,
+      "step": 3410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2335462371563455e-05,
+      "loss": 2.9266,
+      "step": 3412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.233083402758493e-05,
+      "loss": 1.5224,
+      "step": 3414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2326205683606405e-05,
+      "loss": 0.0123,
+      "step": 3416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2321577339627884e-05,
+      "loss": 1.7313,
+      "step": 3418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2316948995649356e-05,
+      "loss": 0.7637,
+      "step": 3420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2312320651670835e-05,
+      "loss": 4.1393,
+      "step": 3422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.230769230769231e-05,
+      "loss": 2.2686,
+      "step": 3424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2303063963713786e-05,
+      "loss": 0.0369,
+      "step": 3426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.229843561973526e-05,
+      "loss": 3.6006,
+      "step": 3428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.229380727575674e-05,
+      "loss": 0.602,
+      "step": 3430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.228917893177821e-05,
+      "loss": 1.6638,
+      "step": 3432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.228455058779969e-05,
+      "loss": 1.8726,
+      "step": 3434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.227992224382116e-05,
+      "loss": 0.0104,
+      "step": 3436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.227529389984264e-05,
+      "loss": 0.6932,
+      "step": 3438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.227066555586412e-05,
+      "loss": 4.7155,
+      "step": 3440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.226603721188559e-05,
+      "loss": 3.8274,
+      "step": 3442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.226140886790707e-05,
+      "loss": 4.0874,
+      "step": 3444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.225678052392854e-05,
+      "loss": 7.9174,
+      "step": 3446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.225215217995002e-05,
+      "loss": 7.3444,
+      "step": 3448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.224752383597149e-05,
+      "loss": 3.3625,
+      "step": 3450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.224289549199297e-05,
+      "loss": 3.9635,
+      "step": 3452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.223826714801444e-05,
+      "loss": 5.6324,
+      "step": 3454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.223363880403592e-05,
+      "loss": 2.6101,
+      "step": 3456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.222901046005739e-05,
+      "loss": 2.7896,
+      "step": 3458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.222438211607887e-05,
+      "loss": 0.6936,
+      "step": 3460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2219753772100344e-05,
+      "loss": 1.3427,
+      "step": 3462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.221512542812182e-05,
+      "loss": 1.1458,
+      "step": 3464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2210497084143294e-05,
+      "loss": 1.0104,
+      "step": 3466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.220586874016477e-05,
+      "loss": 2.0883,
+      "step": 3468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2201240396186245e-05,
+      "loss": 4.5542,
+      "step": 3470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2196612052207724e-05,
+      "loss": 0.8578,
+      "step": 3472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2191983708229196e-05,
+      "loss": 1.1,
+      "step": 3474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2187355364250675e-05,
+      "loss": 2.0326,
+      "step": 3476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.218272702027215e-05,
+      "loss": 2.0999,
+      "step": 3478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2178098676293626e-05,
+      "loss": 1.3926,
+      "step": 3480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2173470332315105e-05,
+      "loss": 0.5853,
+      "step": 3482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.216884198833658e-05,
+      "loss": 0.0059,
+      "step": 3484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2164213644358055e-05,
+      "loss": 3.2007,
+      "step": 3486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.215958530037953e-05,
+      "loss": 2.5404,
+      "step": 3488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2154956956401006e-05,
+      "loss": 5.7034,
+      "step": 3490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.215032861242248e-05,
+      "loss": 0.0066,
+      "step": 3492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.214570026844396e-05,
+      "loss": 2.2369,
+      "step": 3494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.214107192446543e-05,
+      "loss": 1.7382,
+      "step": 3496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.213644358048691e-05,
+      "loss": 1.1092,
+      "step": 3498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.213181523650838e-05,
+      "loss": 0.7164,
+      "step": 3500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.212718689252986e-05,
+      "loss": 4.5722,
+      "step": 3502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.212255854855133e-05,
+      "loss": 2.5489,
+      "step": 3504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.211793020457281e-05,
+      "loss": 7.1832,
+      "step": 3506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.211330186059428e-05,
+      "loss": 4.1472,
+      "step": 3508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2108673516615754e-05,
+      "loss": 1.4145,
+      "step": 3510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.210404517263723e-05,
+      "loss": 2.9268,
+      "step": 3512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2099416828658705e-05,
+      "loss": 7.2677,
+      "step": 3514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2094788484680183e-05,
+      "loss": 1.4109,
+      "step": 3516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2090160140701655e-05,
+      "loss": 4.2786,
+      "step": 3518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2085531796723134e-05,
+      "loss": 5.4463,
+      "step": 3520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2080903452744606e-05,
+      "loss": 3.4037,
+      "step": 3522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2076275108766085e-05,
+      "loss": 1.6205,
+      "step": 3524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.207164676478756e-05,
+      "loss": 0.8215,
+      "step": 3526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2067018420809036e-05,
+      "loss": 1.1193,
+      "step": 3528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.206239007683051e-05,
+      "loss": 0.602,
+      "step": 3530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.205776173285199e-05,
+      "loss": 0.1034,
+      "step": 3532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.205313338887346e-05,
+      "loss": 6.2331,
+      "step": 3534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.204850504489494e-05,
+      "loss": 3.7723,
+      "step": 3536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.204387670091641e-05,
+      "loss": 0.1112,
+      "step": 3538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.203924835693789e-05,
+      "loss": 1.4574,
+      "step": 3540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.203462001295936e-05,
+      "loss": 8.817,
+      "step": 3542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.202999166898084e-05,
+      "loss": 2.4523,
+      "step": 3544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.202536332500232e-05,
+      "loss": 1.8239,
+      "step": 3546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.202073498102379e-05,
+      "loss": 2.7102,
+      "step": 3548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.201610663704527e-05,
+      "loss": 1.357,
+      "step": 3550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.201147829306674e-05,
+      "loss": 2.3576,
+      "step": 3552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.200684994908822e-05,
+      "loss": 1.0694,
+      "step": 3554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.200222160510969e-05,
+      "loss": 6.537,
+      "step": 3556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.199759326113117e-05,
+      "loss": 1.9346,
+      "step": 3558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.199296491715264e-05,
+      "loss": 0.714,
+      "step": 3560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.198833657317412e-05,
+      "loss": 2.4289,
+      "step": 3562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1983708229195594e-05,
+      "loss": 1.2738,
+      "step": 3564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.197907988521707e-05,
+      "loss": 2.4684,
+      "step": 3566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1974451541238544e-05,
+      "loss": 2.5662,
+      "step": 3568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.196982319726002e-05,
+      "loss": 2.4683,
+      "step": 3570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1965194853281495e-05,
+      "loss": 3.6421,
+      "step": 3572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1960566509302974e-05,
+      "loss": 0.7933,
+      "step": 3574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1955938165324446e-05,
+      "loss": 2.0745,
+      "step": 3576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1951309821345925e-05,
+      "loss": 0.6784,
+      "step": 3578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.19466814773674e-05,
+      "loss": 0.868,
+      "step": 3580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1942053133388876e-05,
+      "loss": 4.4735,
+      "step": 3582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.193742478941035e-05,
+      "loss": 3.4782,
+      "step": 3584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.193279644543183e-05,
+      "loss": 2.4448,
+      "step": 3586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1928168101453305e-05,
+      "loss": 0.8041,
+      "step": 3588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.192353975747478e-05,
+      "loss": 3.2576,
+      "step": 3590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1918911413496256e-05,
+      "loss": 1.747,
+      "step": 3592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.191428306951773e-05,
+      "loss": 1.2863,
+      "step": 3594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.190965472553921e-05,
+      "loss": 4.2613,
+      "step": 3596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.190502638156068e-05,
+      "loss": 0.8291,
+      "step": 3598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.190039803758216e-05,
+      "loss": 0.0146,
+      "step": 3600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.189576969360363e-05,
+      "loss": 2.3546,
+      "step": 3602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.189114134962511e-05,
+      "loss": 1.4018,
+      "step": 3604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.188651300564658e-05,
+      "loss": 2.0871,
+      "step": 3606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.188188466166806e-05,
+      "loss": 4.8774,
+      "step": 3608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.187725631768953e-05,
+      "loss": 0.0071,
+      "step": 3610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.187262797371101e-05,
+      "loss": 1.1932,
+      "step": 3612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.186799962973248e-05,
+      "loss": 3.895,
+      "step": 3614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.186337128575396e-05,
+      "loss": 6.5203,
+      "step": 3616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1858742941775433e-05,
+      "loss": 2.5641,
+      "step": 3618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.185411459779691e-05,
+      "loss": 3.6787,
+      "step": 3620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1849486253818384e-05,
+      "loss": 1.2298,
+      "step": 3622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.184485790983986e-05,
+      "loss": 2.7474,
+      "step": 3624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.184022956586134e-05,
+      "loss": 0.588,
+      "step": 3626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1835601221882814e-05,
+      "loss": 1.941,
+      "step": 3628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.183097287790429e-05,
+      "loss": 0.2898,
+      "step": 3630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1826344533925765e-05,
+      "loss": 2.6456,
+      "step": 3632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1821716189947244e-05,
+      "loss": 0.1782,
+      "step": 3634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1817087845968716e-05,
+      "loss": 3.256,
+      "step": 3636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1812459501990194e-05,
+      "loss": 5.2607,
+      "step": 3638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1807831158011667e-05,
+      "loss": 0.0001,
+      "step": 3640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1803202814033145e-05,
+      "loss": 5.2252,
+      "step": 3642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.179857447005462e-05,
+      "loss": 2.5432,
+      "step": 3644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1793946126076096e-05,
+      "loss": 1.1664,
+      "step": 3646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.178931778209757e-05,
+      "loss": 0.8417,
+      "step": 3648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.178468943811905e-05,
+      "loss": 0.7782,
+      "step": 3650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.178006109414052e-05,
+      "loss": 1.3184,
+      "step": 3652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.177543275016199e-05,
+      "loss": 3.1738,
+      "step": 3654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.177080440618347e-05,
+      "loss": 3.2281,
+      "step": 3656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.176617606220494e-05,
+      "loss": 0.5099,
+      "step": 3658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.176154771822642e-05,
+      "loss": 1.7614,
+      "step": 3660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.175691937424789e-05,
+      "loss": 1.164,
+      "step": 3662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.175229103026937e-05,
+      "loss": 0.1932,
+      "step": 3664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1747662686290844e-05,
+      "loss": 0.1416,
+      "step": 3666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.174303434231232e-05,
+      "loss": 0.8173,
+      "step": 3668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1738405998333795e-05,
+      "loss": 2.5322,
+      "step": 3670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.173377765435527e-05,
+      "loss": 0.6402,
+      "step": 3672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1729149310376745e-05,
+      "loss": 0.0021,
+      "step": 3674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1724520966398224e-05,
+      "loss": 0.5181,
+      "step": 3676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1719892622419696e-05,
+      "loss": 1.6885,
+      "step": 3678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1715264278441175e-05,
+      "loss": 0.3627,
+      "step": 3680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.171063593446265e-05,
+      "loss": 2.0161,
+      "step": 3682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1706007590484126e-05,
+      "loss": 5.0827,
+      "step": 3684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.17013792465056e-05,
+      "loss": 5.7841,
+      "step": 3686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.169675090252708e-05,
+      "loss": 6.1674,
+      "step": 3688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.169212255854855e-05,
+      "loss": 4.6215,
+      "step": 3690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.168749421457003e-05,
+      "loss": 2.2734,
+      "step": 3692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1682865870591506e-05,
+      "loss": 2.8671,
+      "step": 3694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.167823752661298e-05,
+      "loss": 0.0236,
+      "step": 3696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.167360918263446e-05,
+      "loss": 0.0064,
+      "step": 3698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.166898083865593e-05,
+      "loss": 5.0424,
+      "step": 3700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.166435249467741e-05,
+      "loss": 7.3294,
+      "step": 3702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.165972415069888e-05,
+      "loss": 1.7349,
+      "step": 3704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.165509580672036e-05,
+      "loss": 2.1063,
+      "step": 3706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.165046746274183e-05,
+      "loss": 0.4167,
+      "step": 3708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.164583911876331e-05,
+      "loss": 3.1664,
+      "step": 3710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.164121077478478e-05,
+      "loss": 1.1138,
+      "step": 3712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.163658243080626e-05,
+      "loss": 0.4404,
+      "step": 3714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.163195408682773e-05,
+      "loss": 1.8831,
+      "step": 3716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.162732574284921e-05,
+      "loss": 1.718,
+      "step": 3718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1622697398870684e-05,
+      "loss": 0.0035,
+      "step": 3720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.161806905489216e-05,
+      "loss": 0.1341,
+      "step": 3722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1613440710913634e-05,
+      "loss": 4.1329,
+      "step": 3724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.160881236693511e-05,
+      "loss": 3.58,
+      "step": 3726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1604184022956585e-05,
+      "loss": 3.091,
+      "step": 3728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1599555678978064e-05,
+      "loss": 1.8734,
+      "step": 3730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1594927334999536e-05,
+      "loss": 2.215,
+      "step": 3732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1590298991021015e-05,
+      "loss": 3.293,
+      "step": 3734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1585670647042494e-05,
+      "loss": 1.3867,
+      "step": 3736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1581042303063966e-05,
+      "loss": 2.058,
+      "step": 3738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1576413959085445e-05,
+      "loss": 4.2126,
+      "step": 3740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1571785615106917e-05,
+      "loss": 2.0042,
+      "step": 3742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1567157271128395e-05,
+      "loss": 2.607,
+      "step": 3744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.156252892714987e-05,
+      "loss": 2.3221,
+      "step": 3746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1557900583171346e-05,
+      "loss": 3.3707,
+      "step": 3748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.155327223919282e-05,
+      "loss": 3.2625,
+      "step": 3750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.15486438952143e-05,
+      "loss": 2.3737,
+      "step": 3752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.154401555123577e-05,
+      "loss": 0.385,
+      "step": 3754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.153938720725725e-05,
+      "loss": 1.3596,
+      "step": 3756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.153475886327872e-05,
+      "loss": 2.6154,
+      "step": 3758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.15301305193002e-05,
+      "loss": 0.7361,
+      "step": 3760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.152550217532167e-05,
+      "loss": 1.0208,
+      "step": 3762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.152087383134315e-05,
+      "loss": 3.1949,
+      "step": 3764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.151624548736462e-05,
+      "loss": 0.0055,
+      "step": 3766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.15116171433861e-05,
+      "loss": 4.3333,
+      "step": 3768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.150698879940757e-05,
+      "loss": 3.7332,
+      "step": 3770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.150236045542905e-05,
+      "loss": 1.3496,
+      "step": 3772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.149773211145053e-05,
+      "loss": 2.2948,
+      "step": 3774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1493103767472e-05,
+      "loss": 2.3364,
+      "step": 3776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.148847542349348e-05,
+      "loss": 0.7023,
+      "step": 3778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.148384707951495e-05,
+      "loss": 0.9714,
+      "step": 3780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.147921873553643e-05,
+      "loss": 1.1263,
+      "step": 3782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1474590391557904e-05,
+      "loss": 3.7788,
+      "step": 3784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.146996204757938e-05,
+      "loss": 4.536,
+      "step": 3786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1465333703600855e-05,
+      "loss": 0.3711,
+      "step": 3788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1460705359622334e-05,
+      "loss": 3.9316,
+      "step": 3790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1456077015643806e-05,
+      "loss": 0.9738,
+      "step": 3792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1451448671665284e-05,
+      "loss": 0.0512,
+      "step": 3794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1446820327686756e-05,
+      "loss": 0.3861,
+      "step": 3796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.144219198370823e-05,
+      "loss": 2.7784,
+      "step": 3798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.143756363972971e-05,
+      "loss": 2.0859,
+      "step": 3800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.143293529575118e-05,
+      "loss": 3.0655,
+      "step": 3802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.142830695177266e-05,
+      "loss": 5.0659,
+      "step": 3804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.142367860779413e-05,
+      "loss": 0.1243,
+      "step": 3806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.141905026381561e-05,
+      "loss": 1.4189,
+      "step": 3808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.141442191983708e-05,
+      "loss": 2.347,
+      "step": 3810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.140979357585856e-05,
+      "loss": 2.277,
+      "step": 3812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.140516523188003e-05,
+      "loss": 3.4847,
+      "step": 3814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.140053688790151e-05,
+      "loss": 0.0679,
+      "step": 3816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.139590854392298e-05,
+      "loss": 1.6648,
+      "step": 3818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.139128019994446e-05,
+      "loss": 4.5147,
+      "step": 3820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1386651855965934e-05,
+      "loss": 1.7526,
+      "step": 3822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.138202351198741e-05,
+      "loss": 1.8708,
+      "step": 3824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1377395168008884e-05,
+      "loss": 3.5462,
+      "step": 3826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.137276682403036e-05,
+      "loss": 2.906,
+      "step": 3828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1368138480051835e-05,
+      "loss": 6.6052,
+      "step": 3830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1363510136073314e-05,
+      "loss": 3.0927,
+      "step": 3832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1358881792094786e-05,
+      "loss": 0.5694,
+      "step": 3834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1354253448116265e-05,
+      "loss": 2.3637,
+      "step": 3836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.134962510413774e-05,
+      "loss": 0.0285,
+      "step": 3838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1344996760159216e-05,
+      "loss": 2.2627,
+      "step": 3840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1340368416180695e-05,
+      "loss": 7.042,
+      "step": 3842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1335740072202167e-05,
+      "loss": 0.7975,
+      "step": 3844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1331111728223645e-05,
+      "loss": 4.4457,
+      "step": 3846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.132648338424512e-05,
+      "loss": 0.1089,
+      "step": 3848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1321855040266596e-05,
+      "loss": 4.7459,
+      "step": 3850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.131722669628807e-05,
+      "loss": 3.1986,
+      "step": 3852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.131259835230955e-05,
+      "loss": 0.0668,
+      "step": 3854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.130797000833102e-05,
+      "loss": 1.4696,
+      "step": 3856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.13033416643525e-05,
+      "loss": 0.0704,
+      "step": 3858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.129871332037397e-05,
+      "loss": 1.3157,
+      "step": 3860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.129408497639545e-05,
+      "loss": 3.2024,
+      "step": 3862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.128945663241692e-05,
+      "loss": 5.3763,
+      "step": 3864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12848282884384e-05,
+      "loss": 3.1467,
+      "step": 3866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.128019994445987e-05,
+      "loss": 1.8381,
+      "step": 3868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.127557160048135e-05,
+      "loss": 5.4926,
+      "step": 3870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.127094325650282e-05,
+      "loss": 1.4235,
+      "step": 3872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12663149125243e-05,
+      "loss": 2.2867,
+      "step": 3874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1261686568545773e-05,
+      "loss": 2.2757,
+      "step": 3876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.125705822456725e-05,
+      "loss": 0.3781,
+      "step": 3878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.125242988058873e-05,
+      "loss": 2.9354,
+      "step": 3880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12478015366102e-05,
+      "loss": 5.6954,
+      "step": 3882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.124317319263168e-05,
+      "loss": 1.1827,
+      "step": 3884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1238544848653154e-05,
+      "loss": 2.0504,
+      "step": 3886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.123391650467463e-05,
+      "loss": 3.7537,
+      "step": 3888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1229288160696105e-05,
+      "loss": 2.0449,
+      "step": 3890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1224659816717584e-05,
+      "loss": 2.7653,
+      "step": 3892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1220031472739056e-05,
+      "loss": 1.9298,
+      "step": 3894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1215403128760534e-05,
+      "loss": 0.1034,
+      "step": 3896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1210774784782006e-05,
+      "loss": 1.6294,
+      "step": 3898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1206146440803485e-05,
+      "loss": 5.753,
+      "step": 3900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.120151809682496e-05,
+      "loss": 4.9546,
+      "step": 3902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1196889752846436e-05,
+      "loss": 0.4539,
+      "step": 3904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.119226140886791e-05,
+      "loss": 3.1163,
+      "step": 3906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.118763306488939e-05,
+      "loss": 2.8255,
+      "step": 3908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.118300472091086e-05,
+      "loss": 0.5978,
+      "step": 3910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.117837637693234e-05,
+      "loss": 2.3276,
+      "step": 3912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.117374803295381e-05,
+      "loss": 0.1532,
+      "step": 3914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.116911968897529e-05,
+      "loss": 1.2292,
+      "step": 3916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.116449134499676e-05,
+      "loss": 7.4231,
+      "step": 3918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.115986300101824e-05,
+      "loss": 0.3868,
+      "step": 3920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.115523465703972e-05,
+      "loss": 2.4503,
+      "step": 3922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.115060631306119e-05,
+      "loss": 1.8351,
+      "step": 3924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.114597796908267e-05,
+      "loss": 1.9912,
+      "step": 3926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.114134962510414e-05,
+      "loss": 4.1159,
+      "step": 3928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.113672128112562e-05,
+      "loss": 2.1446,
+      "step": 3930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.113209293714709e-05,
+      "loss": 3.7215,
+      "step": 3932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.112746459316857e-05,
+      "loss": 2.3043,
+      "step": 3934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.112283624919004e-05,
+      "loss": 1.7064,
+      "step": 3936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.111820790521152e-05,
+      "loss": 0.8269,
+      "step": 3938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1113579561232994e-05,
+      "loss": 0.3153,
+      "step": 3940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.110895121725447e-05,
+      "loss": 3.5597,
+      "step": 3942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.110432287327594e-05,
+      "loss": 1.2079,
+      "step": 3944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.109969452929742e-05,
+      "loss": 0.5997,
+      "step": 3946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1095066185318895e-05,
+      "loss": 0.0103,
+      "step": 3948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.109043784134037e-05,
+      "loss": 12.0473,
+      "step": 3950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1085809497361846e-05,
+      "loss": 3.8525,
+      "step": 3952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.108118115338332e-05,
+      "loss": 7.9891,
+      "step": 3954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.10765528094048e-05,
+      "loss": 5.579,
+      "step": 3956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.107192446542627e-05,
+      "loss": 7.2447,
+      "step": 3958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.106729612144775e-05,
+      "loss": 1.3527,
+      "step": 3960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.106266777746922e-05,
+      "loss": 5.7355,
+      "step": 3962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.10580394334907e-05,
+      "loss": 1.3518,
+      "step": 3964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.105341108951217e-05,
+      "loss": 5.4268,
+      "step": 3966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.104878274553365e-05,
+      "loss": 1.2727,
+      "step": 3968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.104415440155512e-05,
+      "loss": 3.2651,
+      "step": 3970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.10395260575766e-05,
+      "loss": 2.4137,
+      "step": 3972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.103489771359807e-05,
+      "loss": 2.5234,
+      "step": 3974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.103026936961955e-05,
+      "loss": 1.0836,
+      "step": 3976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 3.1837,
+      "step": 3978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.10210126816625e-05,
+      "loss": 3.6531,
+      "step": 3980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1016384337683974e-05,
+      "loss": 2.0475,
+      "step": 3982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.101175599370545e-05,
+      "loss": 2.2145,
+      "step": 3984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.100712764972693e-05,
+      "loss": 1.5748,
+      "step": 3986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.1002499305748404e-05,
+      "loss": 1.3289,
+      "step": 3988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.099787096176988e-05,
+      "loss": 2.0359,
+      "step": 3990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0993242617791355e-05,
+      "loss": 6.8152,
+      "step": 3992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0988614273812834e-05,
+      "loss": 0.3586,
+      "step": 3994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0983985929834306e-05,
+      "loss": 3.4035,
+      "step": 3996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0979357585855784e-05,
+      "loss": 0.1823,
+      "step": 3998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0974729241877256e-05,
+      "loss": 4.0488,
+      "step": 4000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0970100897898735e-05,
+      "loss": 1.9893,
+      "step": 4002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.096547255392021e-05,
+      "loss": 1.4194,
+      "step": 4004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0960844209941686e-05,
+      "loss": 1.3266,
+      "step": 4006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.095621586596316e-05,
+      "loss": 1.321,
+      "step": 4008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.095158752198464e-05,
+      "loss": 1.6558,
+      "step": 4010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.094695917800611e-05,
+      "loss": 2.8843,
+      "step": 4012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.094233083402759e-05,
+      "loss": 5.0278,
+      "step": 4014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.093770249004906e-05,
+      "loss": 0.0043,
+      "step": 4016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.093307414607054e-05,
+      "loss": 0.0009,
+      "step": 4018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.092844580209201e-05,
+      "loss": 3.0047,
+      "step": 4020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.092381745811349e-05,
+      "loss": 4.3861,
+      "step": 4022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.091918911413496e-05,
+      "loss": 1.0815,
+      "step": 4024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.091456077015644e-05,
+      "loss": 0.9404,
+      "step": 4026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.090993242617792e-05,
+      "loss": 3.5903,
+      "step": 4028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.090530408219939e-05,
+      "loss": 4.6203,
+      "step": 4030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.090067573822087e-05,
+      "loss": 2.2168,
+      "step": 4032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.089604739424234e-05,
+      "loss": 1.1086,
+      "step": 4034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.089141905026382e-05,
+      "loss": 1.3785,
+      "step": 4036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.088679070628529e-05,
+      "loss": 1.8327,
+      "step": 4038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.088216236230677e-05,
+      "loss": 5.0805,
+      "step": 4040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0877534018328244e-05,
+      "loss": 1.8962,
+      "step": 4042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.087290567434972e-05,
+      "loss": 3.5338,
+      "step": 4044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0868277330371195e-05,
+      "loss": 2.0095,
+      "step": 4046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0863648986392673e-05,
+      "loss": 0.5762,
+      "step": 4048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0859020642414145e-05,
+      "loss": 2.23,
+      "step": 4050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0854392298435624e-05,
+      "loss": 0.0408,
+      "step": 4052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0849763954457096e-05,
+      "loss": 4.9383,
+      "step": 4054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0845135610478575e-05,
+      "loss": 4.579,
+      "step": 4056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.084050726650005e-05,
+      "loss": 0.2584,
+      "step": 4058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0835878922521526e-05,
+      "loss": 3.4475,
+      "step": 4060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0831250578543e-05,
+      "loss": 1.942,
+      "step": 4062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.082662223456448e-05,
+      "loss": 1.4312,
+      "step": 4064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.082199389058595e-05,
+      "loss": 1.1294,
+      "step": 4066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.081736554660743e-05,
+      "loss": 4.6716,
+      "step": 4068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0812737202628907e-05,
+      "loss": 0.9898,
+      "step": 4070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.080810885865038e-05,
+      "loss": 4.0005,
+      "step": 4072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.080348051467186e-05,
+      "loss": 2.9498,
+      "step": 4074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.079885217069333e-05,
+      "loss": 1.8626,
+      "step": 4076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.079422382671481e-05,
+      "loss": 2.0634,
+      "step": 4078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.078959548273628e-05,
+      "loss": 2.7744,
+      "step": 4080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.078496713875776e-05,
+      "loss": 0.2012,
+      "step": 4082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.078033879477923e-05,
+      "loss": 0.6234,
+      "step": 4084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.077571045080071e-05,
+      "loss": 2.7512,
+      "step": 4086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0771082106822175e-05,
+      "loss": 1.6342,
+      "step": 4088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0766453762843654e-05,
+      "loss": 0.0013,
+      "step": 4090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0761825418865126e-05,
+      "loss": 1.6898,
+      "step": 4092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0757197074886605e-05,
+      "loss": 1.4912,
+      "step": 4094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0752568730908084e-05,
+      "loss": 0.1067,
+      "step": 4096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0747940386929556e-05,
+      "loss": 3.2861,
+      "step": 4098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0743312042951035e-05,
+      "loss": 0.8231,
+      "step": 4100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0738683698972507e-05,
+      "loss": 1.166,
+      "step": 4102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0734055354993985e-05,
+      "loss": 6.3418,
+      "step": 4104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.072942701101546e-05,
+      "loss": 1.4024,
+      "step": 4106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0724798667036936e-05,
+      "loss": 0.1601,
+      "step": 4108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.072017032305841e-05,
+      "loss": 0.3026,
+      "step": 4110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.071554197907989e-05,
+      "loss": 3.2877,
+      "step": 4112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.071091363510136e-05,
+      "loss": 1.5339,
+      "step": 4114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.070628529112284e-05,
+      "loss": 2.8374,
+      "step": 4116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.070165694714431e-05,
+      "loss": 1.7023,
+      "step": 4118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.069702860316579e-05,
+      "loss": 0.3327,
+      "step": 4120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.069240025918726e-05,
+      "loss": 3.2016,
+      "step": 4122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.068777191520874e-05,
+      "loss": 1.9094,
+      "step": 4124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.068314357123021e-05,
+      "loss": 7.8456,
+      "step": 4126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.067851522725169e-05,
+      "loss": 1.5327,
+      "step": 4128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.067388688327316e-05,
+      "loss": 0.7614,
+      "step": 4130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.066925853929464e-05,
+      "loss": 0.0004,
+      "step": 4132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.066463019531612e-05,
+      "loss": 5.4614,
+      "step": 4134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.066000185133759e-05,
+      "loss": 6.818,
+      "step": 4136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.065537350735907e-05,
+      "loss": 4.6532,
+      "step": 4138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.065074516338054e-05,
+      "loss": 0.4316,
+      "step": 4140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.064611681940202e-05,
+      "loss": 1.396,
+      "step": 4142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0641488475423494e-05,
+      "loss": 5.4068,
+      "step": 4144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.063686013144497e-05,
+      "loss": 0.9648,
+      "step": 4146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0632231787466445e-05,
+      "loss": 2.1129,
+      "step": 4148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0627603443487924e-05,
+      "loss": 7.7229,
+      "step": 4150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0622975099509396e-05,
+      "loss": 1.7722,
+      "step": 4152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0618346755530874e-05,
+      "loss": 1.5117,
+      "step": 4154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0613718411552346e-05,
+      "loss": 0.3093,
+      "step": 4156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0609090067573825e-05,
+      "loss": 3.8922,
+      "step": 4158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.06044617235953e-05,
+      "loss": 3.3537,
+      "step": 4160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0599833379616776e-05,
+      "loss": 0.8515,
+      "step": 4162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.059520503563825e-05,
+      "loss": 1.952,
+      "step": 4164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.059057669165973e-05,
+      "loss": 0.4,
+      "step": 4166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.05859483476812e-05,
+      "loss": 0.0022,
+      "step": 4168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.058132000370268e-05,
+      "loss": 0.5044,
+      "step": 4170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.057669165972415e-05,
+      "loss": 1.5347,
+      "step": 4172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.057206331574563e-05,
+      "loss": 1.1235,
+      "step": 4174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.056743497176711e-05,
+      "loss": 2.0561,
+      "step": 4176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.056280662778858e-05,
+      "loss": 0.4242,
+      "step": 4178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.055817828381006e-05,
+      "loss": 1.3928,
+      "step": 4180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.055354993983153e-05,
+      "loss": 0.4701,
+      "step": 4182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.054892159585301e-05,
+      "loss": 3.8047,
+      "step": 4184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.054429325187448e-05,
+      "loss": 4.6918,
+      "step": 4186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.053966490789596e-05,
+      "loss": 6.9842,
+      "step": 4188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.053503656391743e-05,
+      "loss": 1.2547,
+      "step": 4190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.053040821993891e-05,
+      "loss": 4.3598,
+      "step": 4192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.052577987596038e-05,
+      "loss": 0.0019,
+      "step": 4194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.052115153198186e-05,
+      "loss": 2.3273,
+      "step": 4196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0516523188003334e-05,
+      "loss": 2.5301,
+      "step": 4198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.051189484402481e-05,
+      "loss": 0.1585,
+      "step": 4200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0507266500046285e-05,
+      "loss": 4.3762,
+      "step": 4202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.050263815606776e-05,
+      "loss": 3.0751,
+      "step": 4204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0498009812089235e-05,
+      "loss": 0.5771,
+      "step": 4206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0493381468110714e-05,
+      "loss": 2.4967,
+      "step": 4208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0488753124132186e-05,
+      "loss": 1.4627,
+      "step": 4210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0484124780153665e-05,
+      "loss": 3.2215,
+      "step": 4212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0479496436175144e-05,
+      "loss": 2.5603,
+      "step": 4214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0474868092196616e-05,
+      "loss": 0.2096,
+      "step": 4216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0470239748218095e-05,
+      "loss": 1.0174,
+      "step": 4218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.046561140423957e-05,
+      "loss": 1.8722,
+      "step": 4220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0460983060261046e-05,
+      "loss": 0.7786,
+      "step": 4222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.045635471628252e-05,
+      "loss": 2.2448,
+      "step": 4224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0451726372303996e-05,
+      "loss": 4.8662,
+      "step": 4226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.044709802832547e-05,
+      "loss": 4.2928,
+      "step": 4228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.044246968434695e-05,
+      "loss": 3.1654,
+      "step": 4230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.043784134036842e-05,
+      "loss": 1.1447,
+      "step": 4232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.043321299638989e-05,
+      "loss": 6.4044,
+      "step": 4234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.042858465241136e-05,
+      "loss": 0.8323,
+      "step": 4236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.042395630843284e-05,
+      "loss": 2.3208,
+      "step": 4238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.041932796445432e-05,
+      "loss": 1.4892,
+      "step": 4240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.041469962047579e-05,
+      "loss": 1.4713,
+      "step": 4242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.041007127649727e-05,
+      "loss": 0.5545,
+      "step": 4244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0405442932518744e-05,
+      "loss": 0.7224,
+      "step": 4246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.040081458854022e-05,
+      "loss": 3.7163,
+      "step": 4248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0396186244561695e-05,
+      "loss": 3.0758,
+      "step": 4250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0391557900583174e-05,
+      "loss": 5.6511,
+      "step": 4252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0386929556604646e-05,
+      "loss": 3.1082,
+      "step": 4254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0382301212626124e-05,
+      "loss": 0.7766,
+      "step": 4256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0377672868647596e-05,
+      "loss": 4.6442,
+      "step": 4258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0373044524669075e-05,
+      "loss": 0.3585,
+      "step": 4260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.036841618069055e-05,
+      "loss": 6.637,
+      "step": 4262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0363787836712026e-05,
+      "loss": 2.9893,
+      "step": 4264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.03591594927335e-05,
+      "loss": 5.3417,
+      "step": 4266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.035453114875498e-05,
+      "loss": 1.3353,
+      "step": 4268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.034990280477645e-05,
+      "loss": 0.0177,
+      "step": 4270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.034527446079793e-05,
+      "loss": 2.2657,
+      "step": 4272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.03406461168194e-05,
+      "loss": 1.5579,
+      "step": 4274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.033601777284088e-05,
+      "loss": 2.3915,
+      "step": 4276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.033138942886235e-05,
+      "loss": 7.0624,
+      "step": 4278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.032676108488383e-05,
+      "loss": 8.9266,
+      "step": 4280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.032213274090531e-05,
+      "loss": 2.5701,
+      "step": 4282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.031750439692678e-05,
+      "loss": 3.201,
+      "step": 4284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.031287605294826e-05,
+      "loss": 4.4094,
+      "step": 4286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.030824770896973e-05,
+      "loss": 0.5134,
+      "step": 4288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.030361936499121e-05,
+      "loss": 1.792,
+      "step": 4290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.029899102101268e-05,
+      "loss": 0.7249,
+      "step": 4292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.029436267703416e-05,
+      "loss": 3.0788,
+      "step": 4294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.028973433305563e-05,
+      "loss": 2.8909,
+      "step": 4296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.028510598907711e-05,
+      "loss": 2.28,
+      "step": 4298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0280477645098584e-05,
+      "loss": 0.9718,
+      "step": 4300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.027584930112006e-05,
+      "loss": 4.3157,
+      "step": 4302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0271220957141535e-05,
+      "loss": 1.9616,
+      "step": 4304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0266592613163013e-05,
+      "loss": 2.5443,
+      "step": 4306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0261964269184485e-05,
+      "loss": 0.1318,
+      "step": 4308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0257335925205964e-05,
+      "loss": 0.0122,
+      "step": 4310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0252707581227436e-05,
+      "loss": 3.1132,
+      "step": 4312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0248079237248915e-05,
+      "loss": 0.0094,
+      "step": 4314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.024345089327039e-05,
+      "loss": 6.276,
+      "step": 4316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0238822549291866e-05,
+      "loss": 0.0023,
+      "step": 4318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0234194205313345e-05,
+      "loss": 3.5058,
+      "step": 4320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.022956586133482e-05,
+      "loss": 5.4863,
+      "step": 4322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0224937517356296e-05,
+      "loss": 1.3456,
+      "step": 4324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.022030917337777e-05,
+      "loss": 1.2159,
+      "step": 4326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0215680829399246e-05,
+      "loss": 6.2764,
+      "step": 4328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.021105248542072e-05,
+      "loss": 1.7363,
+      "step": 4330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.02064241414422e-05,
+      "loss": 2.4261,
+      "step": 4332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.020179579746367e-05,
+      "loss": 0.3455,
+      "step": 4334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.019716745348515e-05,
+      "loss": 1.1766,
+      "step": 4336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.019253910950662e-05,
+      "loss": 0.0485,
+      "step": 4338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.01879107655281e-05,
+      "loss": 2.3596,
+      "step": 4340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.018328242154957e-05,
+      "loss": 0.0007,
+      "step": 4342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.017865407757105e-05,
+      "loss": 2.3314,
+      "step": 4344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.017402573359252e-05,
+      "loss": 2.0275,
+      "step": 4346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0169397389614e-05,
+      "loss": 2.057,
+      "step": 4348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.016476904563547e-05,
+      "loss": 0.787,
+      "step": 4350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.016014070165695e-05,
+      "loss": 2.3226,
+      "step": 4352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0155512357678424e-05,
+      "loss": 0.3899,
+      "step": 4354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.01508840136999e-05,
+      "loss": 1.1081,
+      "step": 4356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0146255669721374e-05,
+      "loss": 1.9573,
+      "step": 4358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.014162732574285e-05,
+      "loss": 3.5918,
+      "step": 4360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.013699898176433e-05,
+      "loss": 2.4175,
+      "step": 4362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0132370637785804e-05,
+      "loss": 1.1265,
+      "step": 4364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.012774229380728e-05,
+      "loss": 1.9814,
+      "step": 4366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0123113949828755e-05,
+      "loss": 3.7133,
+      "step": 4368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0118485605850234e-05,
+      "loss": 2.6646,
+      "step": 4370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0113857261871706e-05,
+      "loss": 4.5453,
+      "step": 4372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0109228917893185e-05,
+      "loss": 2.2176,
+      "step": 4374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.010460057391466e-05,
+      "loss": 2.1201,
+      "step": 4376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.009997222993613e-05,
+      "loss": 0.6756,
+      "step": 4378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.00953438859576e-05,
+      "loss": 2.1069,
+      "step": 4380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.009071554197908e-05,
+      "loss": 0.4949,
+      "step": 4382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.008608719800055e-05,
+      "loss": 0.2945,
+      "step": 4384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.008145885402203e-05,
+      "loss": 0.0049,
+      "step": 4386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.007683051004351e-05,
+      "loss": 2.5287,
+      "step": 4388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.007220216606498e-05,
+      "loss": 1.9824,
+      "step": 4390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.006757382208646e-05,
+      "loss": 2.1965,
+      "step": 4392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.006294547810793e-05,
+      "loss": 5.2576,
+      "step": 4394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.005831713412941e-05,
+      "loss": 1.4359,
+      "step": 4396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.005368879015088e-05,
+      "loss": 0.7869,
+      "step": 4398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.004906044617236e-05,
+      "loss": 1.6542,
+      "step": 4400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0044432102193834e-05,
+      "loss": 0.0343,
+      "step": 4402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.003980375821531e-05,
+      "loss": 8.9779,
+      "step": 4404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0035175414236785e-05,
+      "loss": 2.3943,
+      "step": 4406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0030547070258263e-05,
+      "loss": 0.5783,
+      "step": 4408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0025918726279735e-05,
+      "loss": 6.561,
+      "step": 4410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0021290382301214e-05,
+      "loss": 0.0152,
+      "step": 4412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0016662038322686e-05,
+      "loss": 0.7434,
+      "step": 4414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0012033694344165e-05,
+      "loss": 4.2037,
+      "step": 4416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.000740535036564e-05,
+      "loss": 2.2369,
+      "step": 4418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0002777006387116e-05,
+      "loss": 5.2959,
+      "step": 4420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.999814866240859e-05,
+      "loss": 1.5798,
+      "step": 4422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.999352031843007e-05,
+      "loss": 1.7281,
+      "step": 4424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.998889197445154e-05,
+      "loss": 2.5788,
+      "step": 4426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.998426363047302e-05,
+      "loss": 0.0967,
+      "step": 4428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9979635286494496e-05,
+      "loss": 2.2562,
+      "step": 4430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997500694251597e-05,
+      "loss": 1.3388,
+      "step": 4432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.997037859853745e-05,
+      "loss": 5.109,
+      "step": 4434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.996575025455892e-05,
+      "loss": 0.4022,
+      "step": 4436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99611219105804e-05,
+      "loss": 0.348,
+      "step": 4438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995649356660187e-05,
+      "loss": 1.5391,
+      "step": 4440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.995186522262335e-05,
+      "loss": 2.2461,
+      "step": 4442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.994723687864482e-05,
+      "loss": 1.9752,
+      "step": 4444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99426085346663e-05,
+      "loss": 0.7827,
+      "step": 4446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993798019068777e-05,
+      "loss": 0.4375,
+      "step": 4448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.993335184670925e-05,
+      "loss": 2.1314,
+      "step": 4450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.992872350273072e-05,
+      "loss": 1.9471,
+      "step": 4452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99240951587522e-05,
+      "loss": 1.1289,
+      "step": 4454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9919466814773674e-05,
+      "loss": 0.2362,
+      "step": 4456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.991483847079515e-05,
+      "loss": 1.528,
+      "step": 4458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9910210126816624e-05,
+      "loss": 6.2223,
+      "step": 4460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.99055817828381e-05,
+      "loss": 3.5753,
+      "step": 4462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9900953438859575e-05,
+      "loss": 0.8897,
+      "step": 4464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9896325094881054e-05,
+      "loss": 3.3783,
+      "step": 4466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.989169675090253e-05,
+      "loss": 3.46,
+      "step": 4468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9887068406924005e-05,
+      "loss": 1.3344,
+      "step": 4470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9882440062945484e-05,
+      "loss": 3.425,
+      "step": 4472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9877811718966956e-05,
+      "loss": 1.7753,
+      "step": 4474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9873183374988435e-05,
+      "loss": 1.6073,
+      "step": 4476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.986855503100991e-05,
+      "loss": 7.4978,
+      "step": 4478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9863926687031385e-05,
+      "loss": 2.2612,
+      "step": 4480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.985929834305286e-05,
+      "loss": 4.1757,
+      "step": 4482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9854669999074336e-05,
+      "loss": 1.2904,
+      "step": 4484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.985004165509581e-05,
+      "loss": 1.5816,
+      "step": 4486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.984541331111729e-05,
+      "loss": 2.2755,
+      "step": 4488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.984078496713876e-05,
+      "loss": 1.2274,
+      "step": 4490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.983615662316024e-05,
+      "loss": 2.6802,
+      "step": 4492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.983152827918171e-05,
+      "loss": 2.8658,
+      "step": 4494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.982689993520319e-05,
+      "loss": 4.0406,
+      "step": 4496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.982227159122466e-05,
+      "loss": 0.5544,
+      "step": 4498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.981764324724614e-05,
+      "loss": 4.6537,
+      "step": 4500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.981301490326761e-05,
+      "loss": 4.3703,
+      "step": 4502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.980838655928909e-05,
+      "loss": 1.4101,
+      "step": 4504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.980375821531056e-05,
+      "loss": 0.5175,
+      "step": 4506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.979912987133204e-05,
+      "loss": 1.8145,
+      "step": 4508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.979450152735352e-05,
+      "loss": 2.4829,
+      "step": 4510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.978987318337499e-05,
+      "loss": 0.9897,
+      "step": 4512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.978524483939647e-05,
+      "loss": 5.1008,
+      "step": 4514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.978061649541794e-05,
+      "loss": 5.1175,
+      "step": 4516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.977598815143942e-05,
+      "loss": 0.8235,
+      "step": 4518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9771359807460894e-05,
+      "loss": 5.4076,
+      "step": 4520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.976673146348237e-05,
+      "loss": 1.9871,
+      "step": 4522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.976210311950384e-05,
+      "loss": 2.0262,
+      "step": 4524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.975747477552532e-05,
+      "loss": 4.065,
+      "step": 4526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.975284643154679e-05,
+      "loss": 0.0108,
+      "step": 4528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.974821808756827e-05,
+      "loss": 0.0038,
+      "step": 4530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 4.4575,
+      "step": 4532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.973896139961122e-05,
+      "loss": 1.2768,
+      "step": 4534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.97343330556327e-05,
+      "loss": 2.3199,
+      "step": 4536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.972970471165417e-05,
+      "loss": 2.0048,
+      "step": 4538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.972507636767565e-05,
+      "loss": 1.9252,
+      "step": 4540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.972044802369712e-05,
+      "loss": 5.9763,
+      "step": 4542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.97158196797186e-05,
+      "loss": 3.3168,
+      "step": 4544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.971119133574007e-05,
+      "loss": 1.6326,
+      "step": 4546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.970656299176155e-05,
+      "loss": 2.0513,
+      "step": 4548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.970193464778302e-05,
+      "loss": 1.9088,
+      "step": 4550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.96973063038045e-05,
+      "loss": 5.2572,
+      "step": 4552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.969267795982597e-05,
+      "loss": 3.707,
+      "step": 4554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.968804961584745e-05,
+      "loss": 0.6561,
+      "step": 4556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9683421271868924e-05,
+      "loss": 2.9832,
+      "step": 4558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.96787929278904e-05,
+      "loss": 5.836,
+      "step": 4560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9674164583911875e-05,
+      "loss": 0.0449,
+      "step": 4562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.966953623993335e-05,
+      "loss": 2.8851,
+      "step": 4564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9664907895954825e-05,
+      "loss": 2.8572,
+      "step": 4566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9660279551976304e-05,
+      "loss": 1.932,
+      "step": 4568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9655651207997776e-05,
+      "loss": 0.9523,
+      "step": 4570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9651022864019255e-05,
+      "loss": 0.5522,
+      "step": 4572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9646394520040734e-05,
+      "loss": 4.3791,
+      "step": 4574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9641766176062206e-05,
+      "loss": 0.509,
+      "step": 4576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9637137832083685e-05,
+      "loss": 4.6404,
+      "step": 4578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.963250948810516e-05,
+      "loss": 1.0693,
+      "step": 4580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9627881144126636e-05,
+      "loss": 1.3206,
+      "step": 4582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.962325280014811e-05,
+      "loss": 0.8808,
+      "step": 4584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9618624456169586e-05,
+      "loss": 2.1969,
+      "step": 4586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.961399611219106e-05,
+      "loss": 5.3653,
+      "step": 4588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.960936776821254e-05,
+      "loss": 3.7456,
+      "step": 4590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.960473942423401e-05,
+      "loss": 3.8856,
+      "step": 4592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.960011108025549e-05,
+      "loss": 0.429,
+      "step": 4594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.959548273627696e-05,
+      "loss": 0.7908,
+      "step": 4596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.959085439229844e-05,
+      "loss": 1.8853,
+      "step": 4598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.958622604831991e-05,
+      "loss": 2.6664,
+      "step": 4600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.958159770434139e-05,
+      "loss": 3.8257,
+      "step": 4602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.957696936036286e-05,
+      "loss": 1.4002,
+      "step": 4604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.957234101638434e-05,
+      "loss": 1.588,
+      "step": 4606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.956771267240581e-05,
+      "loss": 0.5521,
+      "step": 4608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.956308432842729e-05,
+      "loss": 1.0549,
+      "step": 4610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9558455984448764e-05,
+      "loss": 0.1035,
+      "step": 4612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.955382764047024e-05,
+      "loss": 1.3683,
+      "step": 4614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.954919929649172e-05,
+      "loss": 2.3394,
+      "step": 4616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.954457095251319e-05,
+      "loss": 4.025,
+      "step": 4618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.953994260853467e-05,
+      "loss": 4.1844,
+      "step": 4620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9535314264556144e-05,
+      "loss": 0.0213,
+      "step": 4622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.953068592057762e-05,
+      "loss": 1.2124,
+      "step": 4624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9526057576599095e-05,
+      "loss": 1.8971,
+      "step": 4626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9521429232620574e-05,
+      "loss": 2.398,
+      "step": 4628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9516800888642046e-05,
+      "loss": 0.6788,
+      "step": 4630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9512172544663525e-05,
+      "loss": 2.8854,
+      "step": 4632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9507544200684997e-05,
+      "loss": 1.6213,
+      "step": 4634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9502915856706475e-05,
+      "loss": 1.0189,
+      "step": 4636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.949828751272795e-05,
+      "loss": 0.0136,
+      "step": 4638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9493659168749426e-05,
+      "loss": 1.4118,
+      "step": 4640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.94890308247709e-05,
+      "loss": 4.3905,
+      "step": 4642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.948440248079238e-05,
+      "loss": 1.4226,
+      "step": 4644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.947977413681385e-05,
+      "loss": 3.1797,
+      "step": 4646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.947514579283533e-05,
+      "loss": 0.8538,
+      "step": 4648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.94705174488568e-05,
+      "loss": 0.0128,
+      "step": 4650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.946588910487828e-05,
+      "loss": 2.024,
+      "step": 4652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.946126076089975e-05,
+      "loss": 2.9342,
+      "step": 4654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.945663241692123e-05,
+      "loss": 3.1793,
+      "step": 4656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.945200407294271e-05,
+      "loss": 1.088,
+      "step": 4658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.944737572896418e-05,
+      "loss": 1.5399,
+      "step": 4660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.944274738498566e-05,
+      "loss": 0.0477,
+      "step": 4662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.943811904100713e-05,
+      "loss": 2.2955,
+      "step": 4664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.943349069702861e-05,
+      "loss": 2.9965,
+      "step": 4666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9428862353050075e-05,
+      "loss": 4.1841,
+      "step": 4668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9424234009071554e-05,
+      "loss": 1.6437,
+      "step": 4670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9419605665093026e-05,
+      "loss": 0.1889,
+      "step": 4672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9414977321114505e-05,
+      "loss": 2.7675,
+      "step": 4674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.941034897713598e-05,
+      "loss": 2.3276,
+      "step": 4676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9405720633157456e-05,
+      "loss": 3.8957,
+      "step": 4678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9401092289178935e-05,
+      "loss": 3.2478,
+      "step": 4680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.939646394520041e-05,
+      "loss": 2.045,
+      "step": 4682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9391835601221886e-05,
+      "loss": 0.0743,
+      "step": 4684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.938720725724336e-05,
+      "loss": 1.5927,
+      "step": 4686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9382578913264836e-05,
+      "loss": 2.9151,
+      "step": 4688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.937795056928631e-05,
+      "loss": 1.62,
+      "step": 4690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.937332222530779e-05,
+      "loss": 0.1815,
+      "step": 4692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.936869388132926e-05,
+      "loss": 2.3717,
+      "step": 4694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.936406553735074e-05,
+      "loss": 5.7022,
+      "step": 4696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.935943719337221e-05,
+      "loss": 2.7208,
+      "step": 4698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.935480884939369e-05,
+      "loss": 3.6258,
+      "step": 4700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.935018050541516e-05,
+      "loss": 1.1701,
+      "step": 4702
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.934555216143664e-05,
+      "loss": 4.1411,
+      "step": 4704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.934092381745811e-05,
+      "loss": 1.7656,
+      "step": 4706
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.933629547347959e-05,
+      "loss": 4.7991,
+      "step": 4708
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.933166712950106e-05,
+      "loss": 1.0016,
+      "step": 4710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.932703878552254e-05,
+      "loss": 2.9736,
+      "step": 4712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9322410441544014e-05,
+      "loss": 1.3675,
+      "step": 4714
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.931778209756549e-05,
+      "loss": 3.3842,
+      "step": 4716
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9313153753586964e-05,
+      "loss": 3.0452,
+      "step": 4718
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.930852540960844e-05,
+      "loss": 1.6686,
+      "step": 4720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.930389706562992e-05,
+      "loss": 4.0743,
+      "step": 4722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9299268721651394e-05,
+      "loss": 4.9611,
+      "step": 4724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.929464037767287e-05,
+      "loss": 0.0013,
+      "step": 4726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9290012033694345e-05,
+      "loss": 3.8135,
+      "step": 4728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9285383689715824e-05,
+      "loss": 2.6142,
+      "step": 4730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9280755345737296e-05,
+      "loss": 1.6731,
+      "step": 4732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9276127001758775e-05,
+      "loss": 5.0315,
+      "step": 4734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.927149865778025e-05,
+      "loss": 2.7396,
+      "step": 4736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9266870313801725e-05,
+      "loss": 2.6589,
+      "step": 4738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.92622419698232e-05,
+      "loss": 1.0873,
+      "step": 4740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9257613625844676e-05,
+      "loss": 3.6466,
+      "step": 4742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.925298528186615e-05,
+      "loss": 2.4287,
+      "step": 4744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.924835693788763e-05,
+      "loss": 3.077,
+      "step": 4746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.92437285939091e-05,
+      "loss": 1.052,
+      "step": 4748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.923910024993058e-05,
+      "loss": 0.0363,
+      "step": 4750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.923447190595205e-05,
+      "loss": 2.3535,
+      "step": 4752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.922984356197353e-05,
+      "loss": 2.7631,
+      "step": 4754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9225215217995e-05,
+      "loss": 3.0131,
+      "step": 4756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.922058687401648e-05,
+      "loss": 7.7517,
+      "step": 4758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.921595853003795e-05,
+      "loss": 2.4111,
+      "step": 4760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.921133018605943e-05,
+      "loss": 2.5326,
+      "step": 4762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.920670184208091e-05,
+      "loss": 0.0182,
+      "step": 4764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.920207349810238e-05,
+      "loss": 5.6213,
+      "step": 4766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.919744515412386e-05,
+      "loss": 2.2223,
+      "step": 4768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.919281681014533e-05,
+      "loss": 2.5403,
+      "step": 4770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.918818846616681e-05,
+      "loss": 1.0237,
+      "step": 4772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.918356012218828e-05,
+      "loss": 3.13,
+      "step": 4774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.917893177820976e-05,
+      "loss": 6.999,
+      "step": 4776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9174303434231234e-05,
+      "loss": 1.8554,
+      "step": 4778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.916967509025271e-05,
+      "loss": 1.3802,
+      "step": 4780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9165046746274185e-05,
+      "loss": 0.3086,
+      "step": 4782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9160418402295664e-05,
+      "loss": 1.6431,
+      "step": 4784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9155790058317136e-05,
+      "loss": 3.5398,
+      "step": 4786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9151161714338614e-05,
+      "loss": 0.8844,
+      "step": 4788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9146533370360086e-05,
+      "loss": 2.0063,
+      "step": 4790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9141905026381565e-05,
+      "loss": 5.8426,
+      "step": 4792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.913727668240304e-05,
+      "loss": 0.6475,
+      "step": 4794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9132648338424516e-05,
+      "loss": 0.2759,
+      "step": 4796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.912801999444599e-05,
+      "loss": 2.8464,
+      "step": 4798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.912339165046747e-05,
+      "loss": 1.4457,
+      "step": 4800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9118763306488946e-05,
+      "loss": 1.8669,
+      "step": 4802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.911413496251042e-05,
+      "loss": 0.2135,
+      "step": 4804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.91095066185319e-05,
+      "loss": 2.2403,
+      "step": 4806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.910487827455337e-05,
+      "loss": 0.6845,
+      "step": 4808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.910024993057485e-05,
+      "loss": 2.9653,
+      "step": 4810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.909562158659632e-05,
+      "loss": 0.7603,
+      "step": 4812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.909099324261779e-05,
+      "loss": 0.1151,
+      "step": 4814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9086364898639264e-05,
+      "loss": 2.5371,
+      "step": 4816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.908173655466074e-05,
+      "loss": 2.3167,
+      "step": 4818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9077108210682214e-05,
+      "loss": 3.2592,
+      "step": 4820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.907247986670369e-05,
+      "loss": 3.7942,
+      "step": 4822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9067851522725165e-05,
+      "loss": 0.0519,
+      "step": 4824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9063223178746644e-05,
+      "loss": 0.6033,
+      "step": 4826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.905859483476812e-05,
+      "loss": 1.6656,
+      "step": 4828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9053966490789595e-05,
+      "loss": 0.8656,
+      "step": 4830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9049338146811074e-05,
+      "loss": 2.1191,
+      "step": 4832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9044709802832546e-05,
+      "loss": 5.6373,
+      "step": 4834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9040081458854025e-05,
+      "loss": 0.8983,
+      "step": 4836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.90354531148755e-05,
+      "loss": 2.9937,
+      "step": 4838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9030824770896975e-05,
+      "loss": 7.4537,
+      "step": 4840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.902619642691845e-05,
+      "loss": 0.1108,
+      "step": 4842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9021568082939926e-05,
+      "loss": 0.4265,
+      "step": 4844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.90169397389614e-05,
+      "loss": 2.5046,
+      "step": 4846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.901231139498288e-05,
+      "loss": 2.3107,
+      "step": 4848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.900768305100435e-05,
+      "loss": 0.1253,
+      "step": 4850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.900305470702583e-05,
+      "loss": 0.2637,
+      "step": 4852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.89984263630473e-05,
+      "loss": 0.3914,
+      "step": 4854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.899379801906878e-05,
+      "loss": 0.037,
+      "step": 4856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.898916967509025e-05,
+      "loss": 3.1492,
+      "step": 4858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.898454133111173e-05,
+      "loss": 1.905,
+      "step": 4860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.89799129871332e-05,
+      "loss": 0.0708,
+      "step": 4862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.897528464315468e-05,
+      "loss": 2.9641,
+      "step": 4864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.897065629917615e-05,
+      "loss": 1.7082,
+      "step": 4866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.896602795519763e-05,
+      "loss": 7.9043,
+      "step": 4868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.896139961121911e-05,
+      "loss": 2.2043,
+      "step": 4870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.895677126724058e-05,
+      "loss": 5.0196,
+      "step": 4872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.895214292326206e-05,
+      "loss": 0.3016,
+      "step": 4874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.894751457928353e-05,
+      "loss": 4.3036,
+      "step": 4876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.894288623530501e-05,
+      "loss": 2.68,
+      "step": 4878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8938257891326484e-05,
+      "loss": 4.5417,
+      "step": 4880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.893362954734796e-05,
+      "loss": 2.5056,
+      "step": 4882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8929001203369435e-05,
+      "loss": 2.4901,
+      "step": 4884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8924372859390914e-05,
+      "loss": 0.7102,
+      "step": 4886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8919744515412386e-05,
+      "loss": 1.3608,
+      "step": 4888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8915116171433864e-05,
+      "loss": 1.6231,
+      "step": 4890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8910487827455337e-05,
+      "loss": 0.9319,
+      "step": 4892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8905859483476815e-05,
+      "loss": 1.3414,
+      "step": 4894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.890123113949829e-05,
+      "loss": 5.8441,
+      "step": 4896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8896602795519766e-05,
+      "loss": 2.3685,
+      "step": 4898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.889197445154124e-05,
+      "loss": 2.6617,
+      "step": 4900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.888734610756272e-05,
+      "loss": 0.8293,
+      "step": 4902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.888271776358419e-05,
+      "loss": 0.0004,
+      "step": 4904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.887808941960567e-05,
+      "loss": 0.0345,
+      "step": 4906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.887346107562715e-05,
+      "loss": 0.2248,
+      "step": 4908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.886883273164862e-05,
+      "loss": 0.1051,
+      "step": 4910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.88642043876701e-05,
+      "loss": 0.1389,
+      "step": 4912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.885957604369157e-05,
+      "loss": 0.0084,
+      "step": 4914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.885494769971305e-05,
+      "loss": 0.0001,
+      "step": 4916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.885031935573452e-05,
+      "loss": 0.5947,
+      "step": 4918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8845691011756e-05,
+      "loss": 10.7161,
+      "step": 4920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.884106266777747e-05,
+      "loss": 4.4956,
+      "step": 4922
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.883643432379895e-05,
+      "loss": 5.4229,
+      "step": 4924
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.883180597982042e-05,
+      "loss": 9.6197,
+      "step": 4926
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.88271776358419e-05,
+      "loss": 7.5324,
+      "step": 4928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.882254929186337e-05,
+      "loss": 4.4045,
+      "step": 4930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.881792094788485e-05,
+      "loss": 1.107,
+      "step": 4932
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8813292603906324e-05,
+      "loss": 7.7481,
+      "step": 4934
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.88086642599278e-05,
+      "loss": 1.9928,
+      "step": 4936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8804035915949275e-05,
+      "loss": 1.3477,
+      "step": 4938
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8799407571970753e-05,
+      "loss": 2.3497,
+      "step": 4940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8794779227992226e-05,
+      "loss": 2.2197,
+      "step": 4942
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8790150884013704e-05,
+      "loss": 1.3948,
+      "step": 4944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8785522540035176e-05,
+      "loss": 3.4161,
+      "step": 4946
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8780894196056655e-05,
+      "loss": 2.249,
+      "step": 4948
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8776265852078134e-05,
+      "loss": 2.6379,
+      "step": 4950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8771637508099606e-05,
+      "loss": 4.1561,
+      "step": 4952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8767009164121085e-05,
+      "loss": 3.1149,
+      "step": 4954
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.876238082014256e-05,
+      "loss": 0.392,
+      "step": 4956
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.875775247616403e-05,
+      "loss": 1.341,
+      "step": 4958
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.87531241321855e-05,
+      "loss": 0.1616,
+      "step": 4960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.874849578820698e-05,
+      "loss": 2.9244,
+      "step": 4962
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.874386744422845e-05,
+      "loss": 1.9351,
+      "step": 4964
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.873923910024993e-05,
+      "loss": 1.536,
+      "step": 4966
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.87346107562714e-05,
+      "loss": 0.1586,
+      "step": 4968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.872998241229288e-05,
+      "loss": 3.6483,
+      "step": 4970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8725354068314353e-05,
+      "loss": 0.0153,
+      "step": 4972
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.872072572433583e-05,
+      "loss": 1.2945,
+      "step": 4974
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.871609738035731e-05,
+      "loss": 4.8801,
+      "step": 4976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.871146903637878e-05,
+      "loss": 5.0867,
+      "step": 4978
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.870684069240026e-05,
+      "loss": 1.8639,
+      "step": 4980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8702212348421734e-05,
+      "loss": 0.4212,
+      "step": 4982
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.869758400444321e-05,
+      "loss": 4.7299,
+      "step": 4984
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8692955660464685e-05,
+      "loss": 2.7267,
+      "step": 4986
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8688327316486164e-05,
+      "loss": 4.7414,
+      "step": 4988
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8683698972507636e-05,
+      "loss": 2.5881,
+      "step": 4990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8679070628529115e-05,
+      "loss": 0.5599,
+      "step": 4992
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8674442284550587e-05,
+      "loss": 1.5483,
+      "step": 4994
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8669813940572065e-05,
+      "loss": 4.0784,
+      "step": 4996
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.866518559659354e-05,
+      "loss": 0.7261,
+      "step": 4998
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8660557252615016e-05,
+      "loss": 2.8426,
+      "step": 5000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.865592890863649e-05,
+      "loss": 1.5697,
+      "step": 5002
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.865130056465797e-05,
+      "loss": 0.1206,
+      "step": 5004
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.864667222067944e-05,
+      "loss": 0.2807,
+      "step": 5006
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.864204387670092e-05,
+      "loss": 3.3311,
+      "step": 5008
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.863741553272239e-05,
+      "loss": 0.0544,
+      "step": 5010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.863278718874387e-05,
+      "loss": 0.0463,
+      "step": 5012
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.862815884476535e-05,
+      "loss": 6.3057,
+      "step": 5014
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.862353050078682e-05,
+      "loss": 1.5558,
+      "step": 5016
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.86189021568083e-05,
+      "loss": 3.0809,
+      "step": 5018
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.861427381282977e-05,
+      "loss": 0.4249,
+      "step": 5020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.860964546885125e-05,
+      "loss": 1.6821,
+      "step": 5022
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.860501712487272e-05,
+      "loss": 1.2718,
+      "step": 5024
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.86003887808942e-05,
+      "loss": 2.7965,
+      "step": 5026
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.859576043691567e-05,
+      "loss": 6.838,
+      "step": 5028
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.859113209293715e-05,
+      "loss": 2.1504,
+      "step": 5030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.858650374895862e-05,
+      "loss": 0.9382,
+      "step": 5032
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.85818754049801e-05,
+      "loss": 1.3455,
+      "step": 5034
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8577247061001574e-05,
+      "loss": 1.9301,
+      "step": 5036
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.857261871702305e-05,
+      "loss": 4.1787,
+      "step": 5038
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8567990373044525e-05,
+      "loss": 0.995,
+      "step": 5040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8563362029066004e-05,
+      "loss": 0.3902,
+      "step": 5042
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8558733685087476e-05,
+      "loss": 2.2692,
+      "step": 5044
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8554105341108954e-05,
+      "loss": 1.8673,
+      "step": 5046
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8549476997130426e-05,
+      "loss": 4.03,
+      "step": 5048
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8544848653151905e-05,
+      "loss": 1.3055,
+      "step": 5050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.854022030917338e-05,
+      "loss": 4.1863,
+      "step": 5052
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8535591965194856e-05,
+      "loss": 0.5703,
+      "step": 5054
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8530963621216335e-05,
+      "loss": 4.6885,
+      "step": 5056
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.852633527723781e-05,
+      "loss": 1.3242,
+      "step": 5058
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8521706933259286e-05,
+      "loss": 0.1505,
+      "step": 5060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.851707858928076e-05,
+      "loss": 2.5168,
+      "step": 5062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8512450245302237e-05,
+      "loss": 0.5723,
+      "step": 5064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.850782190132371e-05,
+      "loss": 1.6519,
+      "step": 5066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.850319355734519e-05,
+      "loss": 4.9894,
+      "step": 5068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.849856521336666e-05,
+      "loss": 0.1856,
+      "step": 5070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.849393686938814e-05,
+      "loss": 1.777,
+      "step": 5072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.848930852540961e-05,
+      "loss": 2.2732,
+      "step": 5074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.848468018143109e-05,
+      "loss": 2.4011,
+      "step": 5076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.848005183745256e-05,
+      "loss": 3.0266,
+      "step": 5078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.847542349347404e-05,
+      "loss": 1.9835,
+      "step": 5080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.847079514949551e-05,
+      "loss": 5.839,
+      "step": 5082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.846616680551699e-05,
+      "loss": 1.9507,
+      "step": 5084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 2.3814,
+      "step": 5086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.845691011755994e-05,
+      "loss": 0.5134,
+      "step": 5088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8452281773581414e-05,
+      "loss": 2.2499,
+      "step": 5090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.844765342960289e-05,
+      "loss": 2.9633,
+      "step": 5092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8443025085624365e-05,
+      "loss": 2.5803,
+      "step": 5094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.843839674164584e-05,
+      "loss": 2.356,
+      "step": 5096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.843376839766732e-05,
+      "loss": 0.0496,
+      "step": 5098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8429140053688794e-05,
+      "loss": 1.3461,
+      "step": 5100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.842451170971027e-05,
+      "loss": 0.3546,
+      "step": 5102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.841988336573174e-05,
+      "loss": 1.6203,
+      "step": 5104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.841525502175322e-05,
+      "loss": 2.5106,
+      "step": 5106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.841062667777469e-05,
+      "loss": 0.0717,
+      "step": 5108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.840599833379617e-05,
+      "loss": 6.2198,
+      "step": 5110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.840136998981764e-05,
+      "loss": 2.9887,
+      "step": 5112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.839674164583912e-05,
+      "loss": 9.2002,
+      "step": 5114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.839211330186059e-05,
+      "loss": 0.7365,
+      "step": 5116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.838748495788207e-05,
+      "loss": 0.0608,
+      "step": 5118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.838285661390354e-05,
+      "loss": 2.0725,
+      "step": 5120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.837822826992502e-05,
+      "loss": 2.2139,
+      "step": 5122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.83735999259465e-05,
+      "loss": 1.8131,
+      "step": 5124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.836897158196797e-05,
+      "loss": 5.0428,
+      "step": 5126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.836434323798945e-05,
+      "loss": 2.3763,
+      "step": 5128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.835971489401092e-05,
+      "loss": 0.0037,
+      "step": 5130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.83550865500324e-05,
+      "loss": 0.0028,
+      "step": 5132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.835045820605387e-05,
+      "loss": 2.565,
+      "step": 5134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.834582986207535e-05,
+      "loss": 3.7344,
+      "step": 5136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8341201518096824e-05,
+      "loss": 3.4581,
+      "step": 5138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.83365731741183e-05,
+      "loss": 2.9598,
+      "step": 5140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8331944830139775e-05,
+      "loss": 3.2025,
+      "step": 5142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8327316486161254e-05,
+      "loss": 4.3141,
+      "step": 5144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8322688142182726e-05,
+      "loss": 5.3098,
+      "step": 5146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8318059798204204e-05,
+      "loss": 5.6578,
+      "step": 5148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8313431454225676e-05,
+      "loss": 2.0257,
+      "step": 5150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8308803110247155e-05,
+      "loss": 1.9998,
+      "step": 5152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.830417476626863e-05,
+      "loss": 3.8605,
+      "step": 5154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8299546422290106e-05,
+      "loss": 3.4464,
+      "step": 5156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.829491807831158e-05,
+      "loss": 2.5161,
+      "step": 5158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.829028973433306e-05,
+      "loss": 0.3848,
+      "step": 5160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8285661390354536e-05,
+      "loss": 0.3286,
+      "step": 5162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.828103304637601e-05,
+      "loss": 3.7839,
+      "step": 5164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.827640470239749e-05,
+      "loss": 2.2429,
+      "step": 5166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.827177635841896e-05,
+      "loss": 3.4621,
+      "step": 5168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.826714801444044e-05,
+      "loss": 2.5731,
+      "step": 5170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.826251967046191e-05,
+      "loss": 3.2839,
+      "step": 5172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.825789132648339e-05,
+      "loss": 0.2645,
+      "step": 5174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.825326298250486e-05,
+      "loss": 4.986,
+      "step": 5176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.824863463852634e-05,
+      "loss": 2.7931,
+      "step": 5178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.824400629454781e-05,
+      "loss": 1.9307,
+      "step": 5180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.823937795056929e-05,
+      "loss": 7.6953,
+      "step": 5182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.823474960659076e-05,
+      "loss": 1.0336,
+      "step": 5184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.823012126261224e-05,
+      "loss": 2.6635,
+      "step": 5186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.822549291863371e-05,
+      "loss": 2.7224,
+      "step": 5188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.822086457465519e-05,
+      "loss": 4.6668,
+      "step": 5190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8216236230676664e-05,
+      "loss": 2.2009,
+      "step": 5192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.821160788669814e-05,
+      "loss": 0.7188,
+      "step": 5194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8206979542719615e-05,
+      "loss": 3.4383,
+      "step": 5196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8202351198741093e-05,
+      "loss": 6.9004,
+      "step": 5198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8197722854762565e-05,
+      "loss": 1.104,
+      "step": 5200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8193094510784044e-05,
+      "loss": 3.5941,
+      "step": 5202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.818846616680552e-05,
+      "loss": 3.1965,
+      "step": 5204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8183837822826995e-05,
+      "loss": 2.0026,
+      "step": 5206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8179209478848474e-05,
+      "loss": 3.9871,
+      "step": 5208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8174581134869946e-05,
+      "loss": 1.4225,
+      "step": 5210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8169952790891425e-05,
+      "loss": 0.024,
+      "step": 5212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.81653244469129e-05,
+      "loss": 2.7044,
+      "step": 5214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8160696102934376e-05,
+      "loss": 1.3941,
+      "step": 5216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.815606775895585e-05,
+      "loss": 0.271,
+      "step": 5218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8151439414977326e-05,
+      "loss": 0.7559,
+      "step": 5220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.81468110709988e-05,
+      "loss": 0.9346,
+      "step": 5222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.814218272702028e-05,
+      "loss": 1.6737,
+      "step": 5224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.813755438304175e-05,
+      "loss": 0.0022,
+      "step": 5226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.813292603906323e-05,
+      "loss": 2.5481,
+      "step": 5228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.81282976950847e-05,
+      "loss": 0.5465,
+      "step": 5230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.812366935110618e-05,
+      "loss": 2.7756,
+      "step": 5232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.811904100712765e-05,
+      "loss": 5.7632,
+      "step": 5234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.811441266314913e-05,
+      "loss": 2.1915,
+      "step": 5236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.81097843191706e-05,
+      "loss": 3.282,
+      "step": 5238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.810515597519208e-05,
+      "loss": 4.1088,
+      "step": 5240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.810052763121356e-05,
+      "loss": 4.383,
+      "step": 5242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.809589928723503e-05,
+      "loss": 1.5307,
+      "step": 5244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.809127094325651e-05,
+      "loss": 1.9321,
+      "step": 5246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8086642599277976e-05,
+      "loss": 0.3864,
+      "step": 5248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8082014255299454e-05,
+      "loss": 3.9858,
+      "step": 5250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8077385911320926e-05,
+      "loss": 3.2411,
+      "step": 5252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8072757567342405e-05,
+      "loss": 4.6345,
+      "step": 5254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.806812922336388e-05,
+      "loss": 1.2572,
+      "step": 5256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8063500879385356e-05,
+      "loss": 1.6974,
+      "step": 5258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.805887253540683e-05,
+      "loss": 4.5061,
+      "step": 5260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.805424419142831e-05,
+      "loss": 3.5761,
+      "step": 5262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.804961584744978e-05,
+      "loss": 0.6903,
+      "step": 5264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.804498750347126e-05,
+      "loss": 5.0607,
+      "step": 5266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.804035915949274e-05,
+      "loss": 0.3206,
+      "step": 5268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.803573081551421e-05,
+      "loss": 0.9143,
+      "step": 5270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.803110247153569e-05,
+      "loss": 0.1264,
+      "step": 5272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.802647412755716e-05,
+      "loss": 1.9484,
+      "step": 5274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.802184578357864e-05,
+      "loss": 1.5683,
+      "step": 5276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.801721743960011e-05,
+      "loss": 1.4723,
+      "step": 5278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.801258909562159e-05,
+      "loss": 1.4874,
+      "step": 5280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.800796075164306e-05,
+      "loss": 0.9079,
+      "step": 5282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.800333240766454e-05,
+      "loss": 0.6631,
+      "step": 5284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.799870406368601e-05,
+      "loss": 0.9062,
+      "step": 5286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.799407571970749e-05,
+      "loss": 0.6113,
+      "step": 5288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.798944737572896e-05,
+      "loss": 6.3689,
+      "step": 5290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.798481903175044e-05,
+      "loss": 3.1224,
+      "step": 5292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7980190687771914e-05,
+      "loss": 1.6779,
+      "step": 5294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.797556234379339e-05,
+      "loss": 0.0004,
+      "step": 5296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7970933999814865e-05,
+      "loss": 1.2546,
+      "step": 5298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7966305655836343e-05,
+      "loss": 1.4407,
+      "step": 5300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7961677311857815e-05,
+      "loss": 4.6335,
+      "step": 5302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7957048967879294e-05,
+      "loss": 6.8399,
+      "step": 5304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7952420623900766e-05,
+      "loss": 8.094,
+      "step": 5306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7947792279922245e-05,
+      "loss": 2.2177,
+      "step": 5308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7943163935943724e-05,
+      "loss": 5.5943,
+      "step": 5310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7938535591965196e-05,
+      "loss": 0.0181,
+      "step": 5312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7933907247986675e-05,
+      "loss": 1.6041,
+      "step": 5314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.792927890400815e-05,
+      "loss": 1.452,
+      "step": 5316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7924650560029626e-05,
+      "loss": 5.6957,
+      "step": 5318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.79200222160511e-05,
+      "loss": 0.6356,
+      "step": 5320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7915393872072577e-05,
+      "loss": 1.9811,
+      "step": 5322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.791076552809405e-05,
+      "loss": 2.7511,
+      "step": 5324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.790613718411553e-05,
+      "loss": 2.3161,
+      "step": 5326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7901508840137e-05,
+      "loss": 3.8673,
+      "step": 5328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.789688049615848e-05,
+      "loss": 1.6003,
+      "step": 5330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.789225215217995e-05,
+      "loss": 0.9096,
+      "step": 5332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.788762380820143e-05,
+      "loss": 1.1544,
+      "step": 5334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.78829954642229e-05,
+      "loss": 0.0033,
+      "step": 5336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.787836712024438e-05,
+      "loss": 0.9571,
+      "step": 5338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.787373877626585e-05,
+      "loss": 0.001,
+      "step": 5340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.786911043228733e-05,
+      "loss": 1.9321,
+      "step": 5342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.78644820883088e-05,
+      "loss": 0.1667,
+      "step": 5344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.785985374433028e-05,
+      "loss": 6.1003,
+      "step": 5346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7855225400351754e-05,
+      "loss": 1.4425,
+      "step": 5348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.785059705637323e-05,
+      "loss": 2.3592,
+      "step": 5350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.784596871239471e-05,
+      "loss": 1.515,
+      "step": 5352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.784134036841618e-05,
+      "loss": 0.0006,
+      "step": 5354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.783671202443766e-05,
+      "loss": 3.3751,
+      "step": 5356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7832083680459134e-05,
+      "loss": 0.0743,
+      "step": 5358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.782745533648061e-05,
+      "loss": 5.1054,
+      "step": 5360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7822826992502085e-05,
+      "loss": 7.6052,
+      "step": 5362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7818198648523564e-05,
+      "loss": 1.0047,
+      "step": 5364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7813570304545036e-05,
+      "loss": 6.4278,
+      "step": 5366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7808941960566515e-05,
+      "loss": 4.1618,
+      "step": 5368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.780431361658799e-05,
+      "loss": 3.9177,
+      "step": 5370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7799685272609466e-05,
+      "loss": 2.218,
+      "step": 5372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.779505692863094e-05,
+      "loss": 1.1417,
+      "step": 5374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7790428584652416e-05,
+      "loss": 5.7648,
+      "step": 5376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.778580024067389e-05,
+      "loss": 6.1388,
+      "step": 5378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.778117189669537e-05,
+      "loss": 4.8558,
+      "step": 5380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.777654355271684e-05,
+      "loss": 0.4554,
+      "step": 5382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.777191520873832e-05,
+      "loss": 2.8716,
+      "step": 5384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.776728686475979e-05,
+      "loss": 2.2251,
+      "step": 5386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.776265852078127e-05,
+      "loss": 5.503,
+      "step": 5388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.775803017680275e-05,
+      "loss": 0.9463,
+      "step": 5390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.775340183282422e-05,
+      "loss": 1.5516,
+      "step": 5392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.774877348884569e-05,
+      "loss": 0.0038,
+      "step": 5394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7744145144867164e-05,
+      "loss": 4.1381,
+      "step": 5396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.773951680088864e-05,
+      "loss": 0.4716,
+      "step": 5398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7734888456910115e-05,
+      "loss": 4.9654,
+      "step": 5400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7730260112931593e-05,
+      "loss": 2.5872,
+      "step": 5402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7725631768953066e-05,
+      "loss": 3.546,
+      "step": 5404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7721003424974544e-05,
+      "loss": 1.5726,
+      "step": 5406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7716375080996016e-05,
+      "loss": 2.2651,
+      "step": 5408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7711746737017495e-05,
+      "loss": 1.372,
+      "step": 5410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.770711839303897e-05,
+      "loss": 1.7954,
+      "step": 5412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7702490049060446e-05,
+      "loss": 6.6975,
+      "step": 5414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7697861705081925e-05,
+      "loss": 1.4121,
+      "step": 5416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.76932333611034e-05,
+      "loss": 1.1777,
+      "step": 5418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7688605017124876e-05,
+      "loss": 1.8239,
+      "step": 5420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.768397667314635e-05,
+      "loss": 0.5091,
+      "step": 5422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7679348329167827e-05,
+      "loss": 1.5648,
+      "step": 5424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.76747199851893e-05,
+      "loss": 1.0792,
+      "step": 5426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.767009164121078e-05,
+      "loss": 2.9863,
+      "step": 5428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.766546329723225e-05,
+      "loss": 0.2073,
+      "step": 5430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.766083495325373e-05,
+      "loss": 1.5689,
+      "step": 5432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.76562066092752e-05,
+      "loss": 1.5686,
+      "step": 5434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.765157826529668e-05,
+      "loss": 1.1832,
+      "step": 5436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.764694992131815e-05,
+      "loss": 1.4252,
+      "step": 5438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.764232157733963e-05,
+      "loss": 2.125,
+      "step": 5440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.76376932333611e-05,
+      "loss": 3.8125,
+      "step": 5442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.763306488938258e-05,
+      "loss": 0.0066,
+      "step": 5444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.762843654540405e-05,
+      "loss": 0.8037,
+      "step": 5446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.762380820142553e-05,
+      "loss": 0.864,
+      "step": 5448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7619179857447004e-05,
+      "loss": 0.4955,
+      "step": 5450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.761455151346848e-05,
+      "loss": 2.5811,
+      "step": 5452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7609923169489955e-05,
+      "loss": 2.1034,
+      "step": 5454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.760529482551143e-05,
+      "loss": 1.5487,
+      "step": 5456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.760066648153291e-05,
+      "loss": 4.8437,
+      "step": 5458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7596038137554384e-05,
+      "loss": 2.5366,
+      "step": 5460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.759140979357586e-05,
+      "loss": 3.8967,
+      "step": 5462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7586781449597335e-05,
+      "loss": 0.4815,
+      "step": 5464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7582153105618814e-05,
+      "loss": 1.585,
+      "step": 5466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7577524761640286e-05,
+      "loss": 1.2285,
+      "step": 5468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7572896417661765e-05,
+      "loss": 4.3866,
+      "step": 5470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.756826807368324e-05,
+      "loss": 0.0037,
+      "step": 5472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7563639729704716e-05,
+      "loss": 4.1452,
+      "step": 5474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.755901138572619e-05,
+      "loss": 0.1319,
+      "step": 5476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7554383041747666e-05,
+      "loss": 3.4078,
+      "step": 5478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.754975469776914e-05,
+      "loss": 3.2085,
+      "step": 5480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.754512635379062e-05,
+      "loss": 9.0038,
+      "step": 5482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.754049800981209e-05,
+      "loss": 4.3999,
+      "step": 5484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.753586966583357e-05,
+      "loss": 5.9237,
+      "step": 5486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.753124132185504e-05,
+      "loss": 3.5989,
+      "step": 5488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.752661297787652e-05,
+      "loss": 0.7047,
+      "step": 5490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.752198463389799e-05,
+      "loss": 2.0872,
+      "step": 5492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.751735628991947e-05,
+      "loss": 3.2558,
+      "step": 5494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.751272794594095e-05,
+      "loss": 2.7376,
+      "step": 5496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.750809960196242e-05,
+      "loss": 3.8667,
+      "step": 5498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.75034712579839e-05,
+      "loss": 2.8913,
+      "step": 5500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.749884291400537e-05,
+      "loss": 1.4729,
+      "step": 5502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.749421457002685e-05,
+      "loss": 2.8378,
+      "step": 5504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.748958622604832e-05,
+      "loss": 3.0792,
+      "step": 5506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.74849578820698e-05,
+      "loss": 1.2205,
+      "step": 5508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.748032953809127e-05,
+      "loss": 2.6787,
+      "step": 5510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.747570119411275e-05,
+      "loss": 2.3314,
+      "step": 5512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7471072850134224e-05,
+      "loss": 1.2626,
+      "step": 5514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.74664445061557e-05,
+      "loss": 2.5572,
+      "step": 5516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7461816162177175e-05,
+      "loss": 2.3761,
+      "step": 5518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7457187818198654e-05,
+      "loss": 3.2702,
+      "step": 5520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7452559474220126e-05,
+      "loss": 1.7202,
+      "step": 5522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7447931130241605e-05,
+      "loss": 0.1645,
+      "step": 5524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7443302786263077e-05,
+      "loss": 1.6948,
+      "step": 5526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7438674442284555e-05,
+      "loss": 0.9105,
+      "step": 5528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.743404609830603e-05,
+      "loss": 1.1298,
+      "step": 5530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7429417754327506e-05,
+      "loss": 0.0021,
+      "step": 5532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.742478941034898e-05,
+      "loss": 0.3627,
+      "step": 5534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.742016106637046e-05,
+      "loss": 2.2107,
+      "step": 5536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.741553272239193e-05,
+      "loss": 1.3065,
+      "step": 5538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.74109043784134e-05,
+      "loss": 1.5029,
+      "step": 5540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.740627603443488e-05,
+      "loss": 0.876,
+      "step": 5542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.740164769045635e-05,
+      "loss": 0.4546,
+      "step": 5544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.739701934647783e-05,
+      "loss": 6.8861,
+      "step": 5546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.73923910024993e-05,
+      "loss": 1.2284,
+      "step": 5548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.738776265852078e-05,
+      "loss": 0.3435,
+      "step": 5550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7383134314542254e-05,
+      "loss": 1.4988,
+      "step": 5552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.737850597056373e-05,
+      "loss": 0.7403,
+      "step": 5554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7373877626585205e-05,
+      "loss": 2.4201,
+      "step": 5556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7369249282606683e-05,
+      "loss": 6.4367,
+      "step": 5558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7364620938628155e-05,
+      "loss": 0.3026,
+      "step": 5560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7359992594649634e-05,
+      "loss": 2.3819,
+      "step": 5562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.735536425067111e-05,
+      "loss": 2.8059,
+      "step": 5564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7350735906692585e-05,
+      "loss": 1.4897,
+      "step": 5566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7346107562714064e-05,
+      "loss": 0.2903,
+      "step": 5568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7341479218735536e-05,
+      "loss": 4.1454,
+      "step": 5570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7336850874757015e-05,
+      "loss": 6.9496,
+      "step": 5572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.733222253077849e-05,
+      "loss": 4.2972,
+      "step": 5574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7327594186799966e-05,
+      "loss": 0.038,
+      "step": 5576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.732296584282144e-05,
+      "loss": 1.9108,
+      "step": 5578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7318337498842916e-05,
+      "loss": 0.0217,
+      "step": 5580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.731370915486439e-05,
+      "loss": 1.3793,
+      "step": 5582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.730908081088587e-05,
+      "loss": 0.0888,
+      "step": 5584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.730445246690734e-05,
+      "loss": 1.2839,
+      "step": 5586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.729982412292882e-05,
+      "loss": 0.9168,
+      "step": 5588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.729519577895029e-05,
+      "loss": 1.2055,
+      "step": 5590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.729056743497177e-05,
+      "loss": 3.9692,
+      "step": 5592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.728593909099324e-05,
+      "loss": 0.2553,
+      "step": 5594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.728131074701472e-05,
+      "loss": 5.4677,
+      "step": 5596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.727668240303619e-05,
+      "loss": 0.0024,
+      "step": 5598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.727205405905767e-05,
+      "loss": 1.3315,
+      "step": 5600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.726742571507915e-05,
+      "loss": 0.1544,
+      "step": 5602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.726279737110062e-05,
+      "loss": 0.9973,
+      "step": 5604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.72581690271221e-05,
+      "loss": 3.0374,
+      "step": 5606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.725354068314357e-05,
+      "loss": 2.9196,
+      "step": 5608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.724891233916505e-05,
+      "loss": 1.7169,
+      "step": 5610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.724428399518652e-05,
+      "loss": 0.0318,
+      "step": 5612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7239655651208e-05,
+      "loss": 3.1469,
+      "step": 5614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7235027307229474e-05,
+      "loss": 6.1062,
+      "step": 5616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.723039896325095e-05,
+      "loss": 0.7046,
+      "step": 5618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7225770619272425e-05,
+      "loss": 2.917,
+      "step": 5620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7221142275293904e-05,
+      "loss": 6.3644,
+      "step": 5622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7216513931315376e-05,
+      "loss": 3.5582,
+      "step": 5624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7211885587336855e-05,
+      "loss": 0.09,
+      "step": 5626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.720725724335833e-05,
+      "loss": 0.0009,
+      "step": 5628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7202628899379805e-05,
+      "loss": 2.3071,
+      "step": 5630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.719800055540128e-05,
+      "loss": 2.16,
+      "step": 5632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7193372211422756e-05,
+      "loss": 6.2689,
+      "step": 5634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.718874386744423e-05,
+      "loss": 0.3007,
+      "step": 5636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.718411552346571e-05,
+      "loss": 4.3451,
+      "step": 5638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.717948717948718e-05,
+      "loss": 3.9313,
+      "step": 5640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.717485883550866e-05,
+      "loss": 0.0019,
+      "step": 5642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.717023049153014e-05,
+      "loss": 0.2941,
+      "step": 5644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.716560214755161e-05,
+      "loss": 1.3549,
+      "step": 5646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.716097380357309e-05,
+      "loss": 0.8108,
+      "step": 5648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.715634545959456e-05,
+      "loss": 4.2084,
+      "step": 5650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.715171711561604e-05,
+      "loss": 3.3019,
+      "step": 5652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.714708877163751e-05,
+      "loss": 3.6804,
+      "step": 5654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.714246042765899e-05,
+      "loss": 0.8392,
+      "step": 5656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.713783208368046e-05,
+      "loss": 0.2575,
+      "step": 5658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.713320373970194e-05,
+      "loss": 0.5797,
+      "step": 5660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.712857539572341e-05,
+      "loss": 0.0052,
+      "step": 5662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.712394705174489e-05,
+      "loss": 1.2945,
+      "step": 5664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.711931870776636e-05,
+      "loss": 3.1983,
+      "step": 5666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.711469036378784e-05,
+      "loss": 0.0316,
+      "step": 5668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7110062019809314e-05,
+      "loss": 0.0004,
+      "step": 5670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.710543367583079e-05,
+      "loss": 2.8497,
+      "step": 5672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7100805331852265e-05,
+      "loss": 3.1057,
+      "step": 5674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7096176987873744e-05,
+      "loss": 0.7433,
+      "step": 5676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7091548643895216e-05,
+      "loss": 0.9245,
+      "step": 5678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7086920299916694e-05,
+      "loss": 3.9832,
+      "step": 5680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7082291955938166e-05,
+      "loss": 0.5022,
+      "step": 5682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.707766361195964e-05,
+      "loss": 5.4921,
+      "step": 5684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.707303526798112e-05,
+      "loss": 2.8462,
+      "step": 5686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.706840692400259e-05,
+      "loss": 3.0545,
+      "step": 5688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.706377858002407e-05,
+      "loss": 0.0185,
+      "step": 5690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.705915023604554e-05,
+      "loss": 0.2458,
+      "step": 5692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.705452189206702e-05,
+      "loss": 1.6811,
+      "step": 5694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.704989354808849e-05,
+      "loss": 3.7359,
+      "step": 5696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.704526520410997e-05,
+      "loss": 3.3413,
+      "step": 5698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.704063686013144e-05,
+      "loss": 0.0004,
+      "step": 5700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.703600851615292e-05,
+      "loss": 5.206,
+      "step": 5702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.703138017217439e-05,
+      "loss": 1.5071,
+      "step": 5704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.702675182819587e-05,
+      "loss": 2.818,
+      "step": 5706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.7022123484217344e-05,
+      "loss": 0.5471,
+      "step": 5708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.701749514023882e-05,
+      "loss": 5.8081,
+      "step": 5710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.70128667962603e-05,
+      "loss": 5.2874,
+      "step": 5712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.700823845228177e-05,
+      "loss": 0.4417,
+      "step": 5714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.700361010830325e-05,
+      "loss": 6.7978,
+      "step": 5716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6998981764324724e-05,
+      "loss": 4.1572,
+      "step": 5718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.69943534203462e-05,
+      "loss": 5.5485,
+      "step": 5720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6989725076367675e-05,
+      "loss": 3.4657,
+      "step": 5722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6985096732389154e-05,
+      "loss": 1.7109,
+      "step": 5724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6980468388410626e-05,
+      "loss": 7.6438,
+      "step": 5726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6975840044432105e-05,
+      "loss": 3.7325,
+      "step": 5728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.697121170045358e-05,
+      "loss": 1.2219,
+      "step": 5730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6966583356475055e-05,
+      "loss": 1.9663,
+      "step": 5732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.696195501249653e-05,
+      "loss": 3.284,
+      "step": 5734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6957326668518006e-05,
+      "loss": 3.7475,
+      "step": 5736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.695269832453948e-05,
+      "loss": 0.9925,
+      "step": 5738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.694806998056096e-05,
+      "loss": 2.4751,
+      "step": 5740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.694344163658243e-05,
+      "loss": 4.5168,
+      "step": 5742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.693881329260391e-05,
+      "loss": 0.0838,
+      "step": 5744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.693418494862538e-05,
+      "loss": 1.7816,
+      "step": 5746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.692955660464686e-05,
+      "loss": 2.1024,
+      "step": 5748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.692492826066834e-05,
+      "loss": 1.6748,
+      "step": 5750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.692029991668981e-05,
+      "loss": 4.8753,
+      "step": 5752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.691567157271129e-05,
+      "loss": 2.4181,
+      "step": 5754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.691104322873276e-05,
+      "loss": 1.5341,
+      "step": 5756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.690641488475424e-05,
+      "loss": 1.2361,
+      "step": 5758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.690178654077571e-05,
+      "loss": 2.9296,
+      "step": 5760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.689715819679719e-05,
+      "loss": 2.458,
+      "step": 5762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.689252985281866e-05,
+      "loss": 0.6524,
+      "step": 5764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.688790150884014e-05,
+      "loss": 1.2668,
+      "step": 5766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.688327316486161e-05,
+      "loss": 3.8284,
+      "step": 5768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.687864482088309e-05,
+      "loss": 1.1153,
+      "step": 5770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6874016476904564e-05,
+      "loss": 3.1253,
+      "step": 5772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.686938813292604e-05,
+      "loss": 4.7454,
+      "step": 5774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6864759788947515e-05,
+      "loss": 1.214,
+      "step": 5776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6860131444968994e-05,
+      "loss": 4.9028,
+      "step": 5778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6855503100990466e-05,
+      "loss": 1.0108,
+      "step": 5780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6850874757011944e-05,
+      "loss": 2.1585,
+      "step": 5782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6846246413033417e-05,
+      "loss": 1.7023,
+      "step": 5784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6841618069054895e-05,
+      "loss": 2.6818,
+      "step": 5786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.683698972507637e-05,
+      "loss": 3.198,
+      "step": 5788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6832361381097846e-05,
+      "loss": 4.4978,
+      "step": 5790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6827733037119325e-05,
+      "loss": 0.36,
+      "step": 5792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.68231046931408e-05,
+      "loss": 4.1888,
+      "step": 5794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6818476349162276e-05,
+      "loss": 3.7276,
+      "step": 5796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.681384800518375e-05,
+      "loss": 2.5698,
+      "step": 5798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.680921966120523e-05,
+      "loss": 0.8808,
+      "step": 5800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.68045913172267e-05,
+      "loss": 1.1776,
+      "step": 5802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.679996297324818e-05,
+      "loss": 1.738,
+      "step": 5804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.679533462926965e-05,
+      "loss": 2.1031,
+      "step": 5806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.679070628529113e-05,
+      "loss": 1.9786,
+      "step": 5808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.67860779413126e-05,
+      "loss": 1.9495,
+      "step": 5810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.678144959733408e-05,
+      "loss": 1.5391,
+      "step": 5812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.677682125335555e-05,
+      "loss": 0.4247,
+      "step": 5814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.677219290937703e-05,
+      "loss": 2.594,
+      "step": 5816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.67675645653985e-05,
+      "loss": 4.9515,
+      "step": 5818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.676293622141998e-05,
+      "loss": 0.02,
+      "step": 5820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.675830787744145e-05,
+      "loss": 1.0862,
+      "step": 5822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.675367953346293e-05,
+      "loss": 2.0752,
+      "step": 5824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6749051189484404e-05,
+      "loss": 0.3555,
+      "step": 5826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6744422845505876e-05,
+      "loss": 2.5844,
+      "step": 5828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6739794501527355e-05,
+      "loss": 0.9155,
+      "step": 5830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.673516615754883e-05,
+      "loss": 0.4181,
+      "step": 5832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6730537813570306e-05,
+      "loss": 4.7296,
+      "step": 5834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.672590946959178e-05,
+      "loss": 1.6416,
+      "step": 5836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6721281125613256e-05,
+      "loss": 5.7184,
+      "step": 5838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.671665278163473e-05,
+      "loss": 2.388,
+      "step": 5840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.671202443765621e-05,
+      "loss": 5.1771,
+      "step": 5842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.670739609367768e-05,
+      "loss": 1.1083,
+      "step": 5844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.670276774969916e-05,
+      "loss": 1.0888,
+      "step": 5846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.669813940572063e-05,
+      "loss": 5.6697,
+      "step": 5848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.669351106174211e-05,
+      "loss": 1.8407,
+      "step": 5850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.668888271776358e-05,
+      "loss": 2.4514,
+      "step": 5852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.668425437378506e-05,
+      "loss": 0.9198,
+      "step": 5854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.667962602980654e-05,
+      "loss": 0.4436,
+      "step": 5856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.667499768582801e-05,
+      "loss": 1.8852,
+      "step": 5858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.667036934184949e-05,
+      "loss": 0.6295,
+      "step": 5860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.666574099787096e-05,
+      "loss": 6.2802,
+      "step": 5862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.666111265389244e-05,
+      "loss": 1.4778,
+      "step": 5864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.665648430991391e-05,
+      "loss": 2.3817,
+      "step": 5866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.665185596593539e-05,
+      "loss": 3.8138,
+      "step": 5868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.664722762195686e-05,
+      "loss": 0.6826,
+      "step": 5870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.664259927797834e-05,
+      "loss": 1.7768,
+      "step": 5872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6637970933999814e-05,
+      "loss": 0.8487,
+      "step": 5874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.663334259002129e-05,
+      "loss": 5.1746,
+      "step": 5876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6628714246042765e-05,
+      "loss": 0.0661,
+      "step": 5878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6624085902064244e-05,
+      "loss": 0.0008,
+      "step": 5880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6619457558085716e-05,
+      "loss": 4.3587,
+      "step": 5882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6614829214107195e-05,
+      "loss": 0.0065,
+      "step": 5884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6610200870128667e-05,
+      "loss": 0.0649,
+      "step": 5886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6605572526150145e-05,
+      "loss": 2.4656,
+      "step": 5888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.660094418217162e-05,
+      "loss": 3.3962,
+      "step": 5890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6596315838193096e-05,
+      "loss": 7.4604,
+      "step": 5892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.659168749421457e-05,
+      "loss": 0.0284,
+      "step": 5894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.658705915023605e-05,
+      "loss": 5.5076,
+      "step": 5896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6582430806257526e-05,
+      "loss": 1.4567,
+      "step": 5898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6577802462279e-05,
+      "loss": 1.5263,
+      "step": 5900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.657317411830048e-05,
+      "loss": 2.5261,
+      "step": 5902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.656854577432195e-05,
+      "loss": 0.7518,
+      "step": 5904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.656391743034343e-05,
+      "loss": 2.2555,
+      "step": 5906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.65592890863649e-05,
+      "loss": 3.4021,
+      "step": 5908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.655466074238638e-05,
+      "loss": 0.7644,
+      "step": 5910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.655003239840785e-05,
+      "loss": 0.7425,
+      "step": 5912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.654540405442933e-05,
+      "loss": 2.92,
+      "step": 5914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.65407757104508e-05,
+      "loss": 3.7615,
+      "step": 5916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.653614736647228e-05,
+      "loss": 2.7561,
+      "step": 5918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.653151902249375e-05,
+      "loss": 0.5982,
+      "step": 5920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.652689067851523e-05,
+      "loss": 0.7864,
+      "step": 5922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.65222623345367e-05,
+      "loss": 0.3019,
+      "step": 5924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.651763399055818e-05,
+      "loss": 0.2331,
+      "step": 5926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6513005646579654e-05,
+      "loss": 1.7811,
+      "step": 5928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.650837730260113e-05,
+      "loss": 3.9967,
+      "step": 5930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6503748958622605e-05,
+      "loss": 1.1955,
+      "step": 5932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6499120614644084e-05,
+      "loss": 5.445,
+      "step": 5934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.649449227066556e-05,
+      "loss": 0.7575,
+      "step": 5936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6489863926687034e-05,
+      "loss": 2.0712,
+      "step": 5938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.648523558270851e-05,
+      "loss": 4.7086,
+      "step": 5940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6480607238729985e-05,
+      "loss": 0.003,
+      "step": 5942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6475978894751464e-05,
+      "loss": 8.5615,
+      "step": 5944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6471350550772936e-05,
+      "loss": 3.2922,
+      "step": 5946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6466722206794415e-05,
+      "loss": 0.6517,
+      "step": 5948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.646209386281589e-05,
+      "loss": 2.4854,
+      "step": 5950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6457465518837366e-05,
+      "loss": 3.7134,
+      "step": 5952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.645283717485884e-05,
+      "loss": 2.3408,
+      "step": 5954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6448208830880317e-05,
+      "loss": 1.9116,
+      "step": 5956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.644358048690179e-05,
+      "loss": 2.5449,
+      "step": 5958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.643895214292327e-05,
+      "loss": 0.1201,
+      "step": 5960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.643432379894474e-05,
+      "loss": 0.103,
+      "step": 5962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.642969545496622e-05,
+      "loss": 0.0232,
+      "step": 5964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.642506711098769e-05,
+      "loss": 0.9663,
+      "step": 5966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.642043876700917e-05,
+      "loss": 4.4563,
+      "step": 5968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.641581042303064e-05,
+      "loss": 0.0559,
+      "step": 5970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.641118207905212e-05,
+      "loss": 2.3147,
+      "step": 5972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.640655373507359e-05,
+      "loss": 2.0908,
+      "step": 5974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6401925391095064e-05,
+      "loss": 0.0003,
+      "step": 5976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.639729704711654e-05,
+      "loss": 3.7318,
+      "step": 5978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6392668703138015e-05,
+      "loss": 4.9654,
+      "step": 5980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6388040359159494e-05,
+      "loss": 10.8805,
+      "step": 5982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6383412015180966e-05,
+      "loss": 1.9183,
+      "step": 5984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6378783671202445e-05,
+      "loss": 0.0587,
+      "step": 5986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.637415532722392e-05,
+      "loss": 2.9662,
+      "step": 5988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6369526983245395e-05,
+      "loss": 0.9363,
+      "step": 5990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.636489863926687e-05,
+      "loss": 5.6496,
+      "step": 5992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6360270295288346e-05,
+      "loss": 1.0666,
+      "step": 5994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.635564195130982e-05,
+      "loss": 3.6521,
+      "step": 5996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.63510136073313e-05,
+      "loss": 2.9315,
+      "step": 5998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.634638526335277e-05,
+      "loss": 0.0155,
+      "step": 6000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.634175691937425e-05,
+      "loss": 2.5646,
+      "step": 6002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.633712857539573e-05,
+      "loss": 0.3798,
+      "step": 6004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.63325002314172e-05,
+      "loss": 4.6437,
+      "step": 6006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.632787188743868e-05,
+      "loss": 2.8592,
+      "step": 6008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.632324354346015e-05,
+      "loss": 3.5942,
+      "step": 6010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.631861519948163e-05,
+      "loss": 3.8092,
+      "step": 6012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.63139868555031e-05,
+      "loss": 4.0088,
+      "step": 6014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.630935851152458e-05,
+      "loss": 1.0512,
+      "step": 6016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.630473016754605e-05,
+      "loss": 3.2136,
+      "step": 6018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.630010182356753e-05,
+      "loss": 4.5774,
+      "step": 6020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6295473479589e-05,
+      "loss": 1.4297,
+      "step": 6022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.629084513561048e-05,
+      "loss": 2.7349,
+      "step": 6024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.628621679163195e-05,
+      "loss": 0.0743,
+      "step": 6026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.628158844765343e-05,
+      "loss": 1.7092,
+      "step": 6028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6276960103674904e-05,
+      "loss": 3.9974,
+      "step": 6030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.627233175969638e-05,
+      "loss": 0.8572,
+      "step": 6032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6267703415717855e-05,
+      "loss": 0.0522,
+      "step": 6034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6263075071739334e-05,
+      "loss": 2.5119,
+      "step": 6036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6258446727760806e-05,
+      "loss": 2.2725,
+      "step": 6038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6253818383782284e-05,
+      "loss": 3.8868,
+      "step": 6040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6249190039803756e-05,
+      "loss": 0.7213,
+      "step": 6042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6244561695825235e-05,
+      "loss": 3.7715,
+      "step": 6044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6239933351846714e-05,
+      "loss": 1.9637,
+      "step": 6046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6235305007868186e-05,
+      "loss": 0.2706,
+      "step": 6048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6230676663889665e-05,
+      "loss": 3.3656,
+      "step": 6050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.622604831991114e-05,
+      "loss": 1.8178,
+      "step": 6052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6221419975932616e-05,
+      "loss": 1.4188,
+      "step": 6054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.621679163195409e-05,
+      "loss": 6.6426,
+      "step": 6056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.621216328797557e-05,
+      "loss": 0.6246,
+      "step": 6058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.620753494399704e-05,
+      "loss": 2.4584,
+      "step": 6060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.620290660001852e-05,
+      "loss": 1.6974,
+      "step": 6062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.619827825603999e-05,
+      "loss": 0.1049,
+      "step": 6064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.619364991206147e-05,
+      "loss": 0.0168,
+      "step": 6066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.618902156808294e-05,
+      "loss": 5.5314,
+      "step": 6068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.618439322410442e-05,
+      "loss": 3.2488,
+      "step": 6070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.617976488012589e-05,
+      "loss": 2.6756,
+      "step": 6072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.617513653614737e-05,
+      "loss": 5.0494,
+      "step": 6074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.617050819216884e-05,
+      "loss": 0.0158,
+      "step": 6076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.616587984819032e-05,
+      "loss": 3.5959,
+      "step": 6078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.616125150421179e-05,
+      "loss": 1.6582,
+      "step": 6080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.615662316023327e-05,
+      "loss": 4.3228,
+      "step": 6082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.615199481625475e-05,
+      "loss": 1.6021,
+      "step": 6084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.614736647227622e-05,
+      "loss": 1.2701,
+      "step": 6086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.61427381282977e-05,
+      "loss": 1.3065,
+      "step": 6088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6138109784319173e-05,
+      "loss": 1.7874,
+      "step": 6090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.613348144034065e-05,
+      "loss": 1.4222,
+      "step": 6092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6128853096362124e-05,
+      "loss": 1.1869,
+      "step": 6094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.61242247523836e-05,
+      "loss": 0.3538,
+      "step": 6096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6119596408405075e-05,
+      "loss": 3.225,
+      "step": 6098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6114968064426554e-05,
+      "loss": 0.0483,
+      "step": 6100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6110339720448026e-05,
+      "loss": 1.8142,
+      "step": 6102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6105711376469505e-05,
+      "loss": 2.9634,
+      "step": 6104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.610108303249098e-05,
+      "loss": 4.7529,
+      "step": 6106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6096454688512456e-05,
+      "loss": 1.4707,
+      "step": 6108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.609182634453393e-05,
+      "loss": 0.0631,
+      "step": 6110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6087198000555406e-05,
+      "loss": 2.3548,
+      "step": 6112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.608256965657688e-05,
+      "loss": 0.0004,
+      "step": 6114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.607794131259836e-05,
+      "loss": 0.9257,
+      "step": 6116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.607331296861983e-05,
+      "loss": 0.8627,
+      "step": 6118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.60686846246413e-05,
+      "loss": 6.8348,
+      "step": 6120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.606405628066278e-05,
+      "loss": 0.0003,
+      "step": 6122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.605942793668425e-05,
+      "loss": 0.8744,
+      "step": 6124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.605479959270573e-05,
+      "loss": 4.0649,
+      "step": 6126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.60501712487272e-05,
+      "loss": 4.7017,
+      "step": 6128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.604554290474868e-05,
+      "loss": 3.3374,
+      "step": 6130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6040914560770154e-05,
+      "loss": 2.1628,
+      "step": 6132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.603628621679163e-05,
+      "loss": 0.9867,
+      "step": 6134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6031657872813105e-05,
+      "loss": 1.2089,
+      "step": 6136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6027029528834584e-05,
+      "loss": 1.4474,
+      "step": 6138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6022401184856056e-05,
+      "loss": 3.2858,
+      "step": 6140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6017772840877534e-05,
+      "loss": 2.2717,
+      "step": 6142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6013144496899007e-05,
+      "loss": 2.0238,
+      "step": 6144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6008516152920485e-05,
+      "loss": 3.3728,
+      "step": 6146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.600388780894196e-05,
+      "loss": 2.2177,
+      "step": 6148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5999259464963436e-05,
+      "loss": 3.4565,
+      "step": 6150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5994631120984915e-05,
+      "loss": 2.6939,
+      "step": 6152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.599000277700639e-05,
+      "loss": 4.7529,
+      "step": 6154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5985374433027866e-05,
+      "loss": 2.699,
+      "step": 6156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.598074608904934e-05,
+      "loss": 1.7361,
+      "step": 6158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.597611774507082e-05,
+      "loss": 0.9078,
+      "step": 6160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.597148940109229e-05,
+      "loss": 1.8286,
+      "step": 6162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.596686105711377e-05,
+      "loss": 3.6526,
+      "step": 6164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.596223271313524e-05,
+      "loss": 2.3326,
+      "step": 6166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.595760436915672e-05,
+      "loss": 2.8203,
+      "step": 6168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.595297602517819e-05,
+      "loss": 0.1101,
+      "step": 6170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.594834768119967e-05,
+      "loss": 1.3344,
+      "step": 6172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.594371933722114e-05,
+      "loss": 0.0612,
+      "step": 6174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.593909099324262e-05,
+      "loss": 0.0122,
+      "step": 6176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.593446264926409e-05,
+      "loss": 0.2434,
+      "step": 6178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.592983430528557e-05,
+      "loss": 2.1963,
+      "step": 6180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.592520596130704e-05,
+      "loss": 1.9084,
+      "step": 6182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.592057761732852e-05,
+      "loss": 2.9722,
+      "step": 6184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5915949273349994e-05,
+      "loss": 2.6964,
+      "step": 6186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.591132092937147e-05,
+      "loss": 6.7566,
+      "step": 6188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.590669258539295e-05,
+      "loss": 0.0095,
+      "step": 6190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5902064241414423e-05,
+      "loss": 1.7288,
+      "step": 6192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 4.7589,
+      "step": 6194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5892807553457374e-05,
+      "loss": 1.4515,
+      "step": 6196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.588817920947885e-05,
+      "loss": 2.6139,
+      "step": 6198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5883550865500325e-05,
+      "loss": 1.6919,
+      "step": 6200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5878922521521804e-05,
+      "loss": 1.4653,
+      "step": 6202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5874294177543276e-05,
+      "loss": 1.531,
+      "step": 6204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5869665833564755e-05,
+      "loss": 6.292,
+      "step": 6206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.586503748958623e-05,
+      "loss": 0.8069,
+      "step": 6208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5860409145607706e-05,
+      "loss": 3.0635,
+      "step": 6210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.585578080162918e-05,
+      "loss": 2.3639,
+      "step": 6212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5851152457650657e-05,
+      "loss": 5.0818,
+      "step": 6214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.584652411367213e-05,
+      "loss": 2.1696,
+      "step": 6216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.584189576969361e-05,
+      "loss": 0.5575,
+      "step": 6218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.583726742571508e-05,
+      "loss": 5.9583,
+      "step": 6220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.583263908173656e-05,
+      "loss": 1.9208,
+      "step": 6222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.582801073775803e-05,
+      "loss": 2.6501,
+      "step": 6224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.582338239377951e-05,
+      "loss": 2.1868,
+      "step": 6226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.581875404980098e-05,
+      "loss": 1.9882,
+      "step": 6228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.581412570582246e-05,
+      "loss": 0.0597,
+      "step": 6230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.580949736184394e-05,
+      "loss": 1.6396,
+      "step": 6232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.580486901786541e-05,
+      "loss": 1.5216,
+      "step": 6234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.580024067388689e-05,
+      "loss": 1.9029,
+      "step": 6236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.579561232990836e-05,
+      "loss": 0.3035,
+      "step": 6238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.579098398592984e-05,
+      "loss": 1.3344,
+      "step": 6240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.578635564195131e-05,
+      "loss": 1.353,
+      "step": 6242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.578172729797279e-05,
+      "loss": 3.1444,
+      "step": 6244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.577709895399426e-05,
+      "loss": 0.9602,
+      "step": 6246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.577247061001574e-05,
+      "loss": 1.8484,
+      "step": 6248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5767842266037214e-05,
+      "loss": 2.581,
+      "step": 6250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.576321392205869e-05,
+      "loss": 2.9517,
+      "step": 6252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5758585578080165e-05,
+      "loss": 5.4905,
+      "step": 6254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5753957234101644e-05,
+      "loss": 4.2167,
+      "step": 6256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5749328890123116e-05,
+      "loss": 2.2642,
+      "step": 6258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5744700546144595e-05,
+      "loss": 4.9888,
+      "step": 6260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.574007220216607e-05,
+      "loss": 1.7525,
+      "step": 6262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.573544385818754e-05,
+      "loss": 5.1173,
+      "step": 6264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.573081551420902e-05,
+      "loss": 0.1202,
+      "step": 6266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.572618717023049e-05,
+      "loss": 1.8679,
+      "step": 6268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.572155882625197e-05,
+      "loss": 6.0863,
+      "step": 6270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.571693048227344e-05,
+      "loss": 2.0076,
+      "step": 6272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.571230213829492e-05,
+      "loss": 0.7637,
+      "step": 6274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.570767379431639e-05,
+      "loss": 1.9778,
+      "step": 6276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.570304545033787e-05,
+      "loss": 0.9807,
+      "step": 6278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.569841710635934e-05,
+      "loss": 1.3302,
+      "step": 6280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.569378876238082e-05,
+      "loss": 3.2368,
+      "step": 6282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.568916041840229e-05,
+      "loss": 5.3624,
+      "step": 6284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.568453207442377e-05,
+      "loss": 5.2366,
+      "step": 6286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5679903730445244e-05,
+      "loss": 0.432,
+      "step": 6288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.567527538646672e-05,
+      "loss": 4.274,
+      "step": 6290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5670647042488195e-05,
+      "loss": 0.0405,
+      "step": 6292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5666018698509674e-05,
+      "loss": 1.0779,
+      "step": 6294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.566139035453115e-05,
+      "loss": 2.5997,
+      "step": 6296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5656762010552624e-05,
+      "loss": 2.7718,
+      "step": 6298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.56521336665741e-05,
+      "loss": 2.3072,
+      "step": 6300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5647505322595575e-05,
+      "loss": 1.613,
+      "step": 6302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5642876978617054e-05,
+      "loss": 3.2601,
+      "step": 6304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5638248634638526e-05,
+      "loss": 1.656,
+      "step": 6306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5633620290660005e-05,
+      "loss": 1.537,
+      "step": 6308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.562899194668148e-05,
+      "loss": 1.8746,
+      "step": 6310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5624363602702956e-05,
+      "loss": 4.1175,
+      "step": 6312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.561973525872443e-05,
+      "loss": 4.0924,
+      "step": 6314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5615106914745907e-05,
+      "loss": 0.2121,
+      "step": 6316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.561047857076738e-05,
+      "loss": 3.9925,
+      "step": 6318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.560585022678886e-05,
+      "loss": 4.606,
+      "step": 6320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.560122188281033e-05,
+      "loss": 0.0094,
+      "step": 6322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.559659353883181e-05,
+      "loss": 0.9487,
+      "step": 6324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.559196519485328e-05,
+      "loss": 2.1643,
+      "step": 6326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.558733685087476e-05,
+      "loss": 2.7325,
+      "step": 6328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.558270850689623e-05,
+      "loss": 2.2619,
+      "step": 6330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.557808016291771e-05,
+      "loss": 8.0316,
+      "step": 6332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.557345181893918e-05,
+      "loss": 4.8703,
+      "step": 6334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.556882347496066e-05,
+      "loss": 4.3353,
+      "step": 6336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.556419513098214e-05,
+      "loss": 2.4057,
+      "step": 6338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.555956678700361e-05,
+      "loss": 3.6493,
+      "step": 6340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.555493844302509e-05,
+      "loss": 0.8267,
+      "step": 6342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.555031009904656e-05,
+      "loss": 5.0896,
+      "step": 6344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.554568175506804e-05,
+      "loss": 3.2058,
+      "step": 6346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.554105341108951e-05,
+      "loss": 3.4696,
+      "step": 6348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.553642506711099e-05,
+      "loss": 6.3969,
+      "step": 6350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5531796723132464e-05,
+      "loss": 1.5527,
+      "step": 6352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.552716837915394e-05,
+      "loss": 1.6831,
+      "step": 6354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5522540035175415e-05,
+      "loss": 0.4649,
+      "step": 6356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5517911691196894e-05,
+      "loss": 0.9041,
+      "step": 6358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5513283347218366e-05,
+      "loss": 1.4439,
+      "step": 6360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5508655003239845e-05,
+      "loss": 4.1808,
+      "step": 6362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.550402665926132e-05,
+      "loss": 0.7698,
+      "step": 6364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5499398315282796e-05,
+      "loss": 0.4028,
+      "step": 6366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.549476997130427e-05,
+      "loss": 0.0617,
+      "step": 6368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5490141627325746e-05,
+      "loss": 2.7425,
+      "step": 6370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.548551328334722e-05,
+      "loss": 3.8886,
+      "step": 6372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.54808849393687e-05,
+      "loss": 4.3806,
+      "step": 6374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.547625659539017e-05,
+      "loss": 2.2566,
+      "step": 6376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.547162825141165e-05,
+      "loss": 0.6849,
+      "step": 6378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.546699990743313e-05,
+      "loss": 2.1014,
+      "step": 6380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.54623715634546e-05,
+      "loss": 0.9316,
+      "step": 6382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.545774321947608e-05,
+      "loss": 2.9032,
+      "step": 6384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.545311487549755e-05,
+      "loss": 2.8254,
+      "step": 6386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.544848653151903e-05,
+      "loss": 5.0328,
+      "step": 6388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.54438581875405e-05,
+      "loss": 4.252,
+      "step": 6390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.543922984356198e-05,
+      "loss": 2.4049,
+      "step": 6392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.543460149958345e-05,
+      "loss": 1.4766,
+      "step": 6394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.542997315560493e-05,
+      "loss": 3.2125,
+      "step": 6396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.54253448116264e-05,
+      "loss": 3.9578,
+      "step": 6398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.542071646764788e-05,
+      "loss": 1.2052,
+      "step": 6400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.541608812366935e-05,
+      "loss": 3.9805,
+      "step": 6402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.541145977969083e-05,
+      "loss": 0.7248,
+      "step": 6404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5406831435712304e-05,
+      "loss": 0.0032,
+      "step": 6406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5402203091733776e-05,
+      "loss": 1.1156,
+      "step": 6408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5397574747755255e-05,
+      "loss": 2.2453,
+      "step": 6410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.539294640377673e-05,
+      "loss": 4.924,
+      "step": 6412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5388318059798206e-05,
+      "loss": 3.3819,
+      "step": 6414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.538368971581968e-05,
+      "loss": 3.3417,
+      "step": 6416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.537906137184116e-05,
+      "loss": 0.619,
+      "step": 6418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.537443302786263e-05,
+      "loss": 2.0827,
+      "step": 6420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.536980468388411e-05,
+      "loss": 2.3757,
+      "step": 6422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.536517633990558e-05,
+      "loss": 2.6508,
+      "step": 6424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.536054799592706e-05,
+      "loss": 0.4813,
+      "step": 6426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.535591965194853e-05,
+      "loss": 1.1806,
+      "step": 6428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.535129130797001e-05,
+      "loss": 2.312,
+      "step": 6430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.534666296399148e-05,
+      "loss": 3.6402,
+      "step": 6432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.534203462001296e-05,
+      "loss": 0.5747,
+      "step": 6434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.533740627603443e-05,
+      "loss": 2.7136,
+      "step": 6436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.533277793205591e-05,
+      "loss": 6.3394,
+      "step": 6438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.532814958807738e-05,
+      "loss": 2.585,
+      "step": 6440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.532352124409886e-05,
+      "loss": 0.2716,
+      "step": 6442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.531889290012034e-05,
+      "loss": 1.9206,
+      "step": 6444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.531426455614181e-05,
+      "loss": 1.2189,
+      "step": 6446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.530963621216329e-05,
+      "loss": 0.3429,
+      "step": 6448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5305007868184763e-05,
+      "loss": 0.899,
+      "step": 6450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.530037952420624e-05,
+      "loss": 1.0293,
+      "step": 6452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5295751180227714e-05,
+      "loss": 1.3007,
+      "step": 6454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.529112283624919e-05,
+      "loss": 6.9037,
+      "step": 6456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5286494492270665e-05,
+      "loss": 0.1019,
+      "step": 6458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5281866148292144e-05,
+      "loss": 3.0572,
+      "step": 6460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5277237804313616e-05,
+      "loss": 0.1321,
+      "step": 6462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5272609460335095e-05,
+      "loss": 1.2291,
+      "step": 6464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.526798111635657e-05,
+      "loss": 5.5503,
+      "step": 6466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5263352772378046e-05,
+      "loss": 2.3046,
+      "step": 6468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.525872442839952e-05,
+      "loss": 0.8816,
+      "step": 6470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5254096084420996e-05,
+      "loss": 1.8612,
+      "step": 6472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.524946774044247e-05,
+      "loss": 1.0378,
+      "step": 6474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.524483939646395e-05,
+      "loss": 4.6728,
+      "step": 6476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.524021105248542e-05,
+      "loss": 3.0004,
+      "step": 6478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.52355827085069e-05,
+      "loss": 2.2216,
+      "step": 6480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.523095436452837e-05,
+      "loss": 2.0644,
+      "step": 6482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.522632602054985e-05,
+      "loss": 2.2373,
+      "step": 6484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.522169767657133e-05,
+      "loss": 1.8305,
+      "step": 6486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.52170693325928e-05,
+      "loss": 5.1282,
+      "step": 6488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.521244098861428e-05,
+      "loss": 2.8125,
+      "step": 6490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.520781264463575e-05,
+      "loss": 2.8627,
+      "step": 6492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.520318430065723e-05,
+      "loss": 0.6792,
+      "step": 6494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.51985559566787e-05,
+      "loss": 0.9878,
+      "step": 6496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.519392761270018e-05,
+      "loss": 0.2892,
+      "step": 6498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.518929926872165e-05,
+      "loss": 0.0783,
+      "step": 6500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.518467092474313e-05,
+      "loss": 5.6691,
+      "step": 6502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.51800425807646e-05,
+      "loss": 0.8315,
+      "step": 6504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.517541423678608e-05,
+      "loss": 3.2097,
+      "step": 6506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5170785892807554e-05,
+      "loss": 3.2165,
+      "step": 6508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.516615754882903e-05,
+      "loss": 0.0331,
+      "step": 6510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5161529204850505e-05,
+      "loss": 0.5675,
+      "step": 6512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5156900860871984e-05,
+      "loss": 1.0426,
+      "step": 6514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5152272516893456e-05,
+      "loss": 0.8278,
+      "step": 6516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5147644172914935e-05,
+      "loss": 1.2118,
+      "step": 6518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.514301582893641e-05,
+      "loss": 5.2994,
+      "step": 6520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5138387484957885e-05,
+      "loss": 2.2023,
+      "step": 6522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5133759140979364e-05,
+      "loss": 2.0936,
+      "step": 6524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5129130797000836e-05,
+      "loss": 0.6253,
+      "step": 6526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5124502453022315e-05,
+      "loss": 4.5296,
+      "step": 6528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.511987410904379e-05,
+      "loss": 2.4657,
+      "step": 6530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5115245765065266e-05,
+      "loss": 4.2719,
+      "step": 6532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.511061742108674e-05,
+      "loss": 0.926,
+      "step": 6534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.510598907710822e-05,
+      "loss": 3.737,
+      "step": 6536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.510136073312969e-05,
+      "loss": 4.6151,
+      "step": 6538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.509673238915117e-05,
+      "loss": 2.7746,
+      "step": 6540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.509210404517264e-05,
+      "loss": 3.4812,
+      "step": 6542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.508747570119412e-05,
+      "loss": 1.5641,
+      "step": 6544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.508284735721559e-05,
+      "loss": 1.6813,
+      "step": 6546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.507821901323707e-05,
+      "loss": 0.4744,
+      "step": 6548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.507359066925854e-05,
+      "loss": 3.4244,
+      "step": 6550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.506896232528002e-05,
+      "loss": 0.1827,
+      "step": 6552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.506433398130149e-05,
+      "loss": 3.9521,
+      "step": 6554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5059705637322964e-05,
+      "loss": 0.0408,
+      "step": 6556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.505507729334444e-05,
+      "loss": 4.701,
+      "step": 6558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5050448949365915e-05,
+      "loss": 4.3409,
+      "step": 6560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5045820605387394e-05,
+      "loss": 3.2009,
+      "step": 6562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5041192261408866e-05,
+      "loss": 1.3537,
+      "step": 6564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5036563917430345e-05,
+      "loss": 1.2877,
+      "step": 6566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.503193557345182e-05,
+      "loss": 2.0135,
+      "step": 6568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5027307229473296e-05,
+      "loss": 2.6924,
+      "step": 6570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.502267888549477e-05,
+      "loss": 0.0741,
+      "step": 6572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5018050541516247e-05,
+      "loss": 2.9057,
+      "step": 6574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.501342219753772e-05,
+      "loss": 0.0343,
+      "step": 6576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.50087938535592e-05,
+      "loss": 2.0718,
+      "step": 6578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.500416550958067e-05,
+      "loss": 1.6488,
+      "step": 6580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.499953716560215e-05,
+      "loss": 4.005,
+      "step": 6582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.499490882162362e-05,
+      "loss": 1.1,
+      "step": 6584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.49902804776451e-05,
+      "loss": 0.0787,
+      "step": 6586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.498565213366657e-05,
+      "loss": 0.2574,
+      "step": 6588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.498102378968805e-05,
+      "loss": 0.2301,
+      "step": 6590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.497639544570953e-05,
+      "loss": 0.582,
+      "step": 6592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4971767101731e-05,
+      "loss": 2.7849,
+      "step": 6594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.496713875775248e-05,
+      "loss": 0.661,
+      "step": 6596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.496251041377395e-05,
+      "loss": 3.1386,
+      "step": 6598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.495788206979543e-05,
+      "loss": 5.8302,
+      "step": 6600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.49532537258169e-05,
+      "loss": 5.7031,
+      "step": 6602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.494862538183838e-05,
+      "loss": 0.7778,
+      "step": 6604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.494399703785985e-05,
+      "loss": 10.9303,
+      "step": 6606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.493936869388133e-05,
+      "loss": 4.0441,
+      "step": 6608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4934740349902804e-05,
+      "loss": 0.7041,
+      "step": 6610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.493011200592428e-05,
+      "loss": 1.3676,
+      "step": 6612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4925483661945755e-05,
+      "loss": 4.3869,
+      "step": 6614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4920855317967234e-05,
+      "loss": 4.3033,
+      "step": 6616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4916226973988706e-05,
+      "loss": 1.6583,
+      "step": 6618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4911598630010185e-05,
+      "loss": 3.9062,
+      "step": 6620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.490697028603166e-05,
+      "loss": 1.0324,
+      "step": 6622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4902341942053136e-05,
+      "loss": 1.6709,
+      "step": 6624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.489771359807461e-05,
+      "loss": 2.0905,
+      "step": 6626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4893085254096086e-05,
+      "loss": 0.4015,
+      "step": 6628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4888456910117565e-05,
+      "loss": 0.3686,
+      "step": 6630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.488382856613904e-05,
+      "loss": 4.5366,
+      "step": 6632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4879200222160516e-05,
+      "loss": 1.696,
+      "step": 6634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.487457187818199e-05,
+      "loss": 0.6729,
+      "step": 6636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.486994353420347e-05,
+      "loss": 0.9021,
+      "step": 6638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.486531519022494e-05,
+      "loss": 1.8651,
+      "step": 6640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.486068684624642e-05,
+      "loss": 5.343,
+      "step": 6642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.485605850226789e-05,
+      "loss": 1.8399,
+      "step": 6644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.485143015828937e-05,
+      "loss": 2.9084,
+      "step": 6646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.484680181431084e-05,
+      "loss": 0.0079,
+      "step": 6648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.484217347033232e-05,
+      "loss": 5.2796,
+      "step": 6650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.483754512635379e-05,
+      "loss": 1.7186,
+      "step": 6652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.483291678237527e-05,
+      "loss": 3.7599,
+      "step": 6654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.482828843839674e-05,
+      "loss": 1.8109,
+      "step": 6656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.482366009441822e-05,
+      "loss": 1.8401,
+      "step": 6658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.481903175043969e-05,
+      "loss": 3.4504,
+      "step": 6660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.481440340646117e-05,
+      "loss": 1.2697,
+      "step": 6662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4809775062482644e-05,
+      "loss": 2.455,
+      "step": 6664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.480514671850412e-05,
+      "loss": 6.1099,
+      "step": 6666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4800518374525595e-05,
+      "loss": 7.3682,
+      "step": 6668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4795890030547074e-05,
+      "loss": 1.1983,
+      "step": 6670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.479126168656855e-05,
+      "loss": 0.5042,
+      "step": 6672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4786633342590025e-05,
+      "loss": 0.0694,
+      "step": 6674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.47820049986115e-05,
+      "loss": 2.5893,
+      "step": 6676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4777376654632975e-05,
+      "loss": 4.3696,
+      "step": 6678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4772748310654454e-05,
+      "loss": 3.769,
+      "step": 6680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4768119966675926e-05,
+      "loss": 1.3759,
+      "step": 6682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4763491622697405e-05,
+      "loss": 4.2688,
+      "step": 6684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.475886327871888e-05,
+      "loss": 1.6184,
+      "step": 6686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4754234934740356e-05,
+      "loss": 1.2228,
+      "step": 6688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.474960659076183e-05,
+      "loss": 0.4686,
+      "step": 6690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.474497824678331e-05,
+      "loss": 3.4913,
+      "step": 6692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.474034990280478e-05,
+      "loss": 0.2598,
+      "step": 6694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.473572155882626e-05,
+      "loss": 2.1474,
+      "step": 6696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.473109321484773e-05,
+      "loss": 3.4515,
+      "step": 6698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.47264648708692e-05,
+      "loss": 0.5865,
+      "step": 6700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.472183652689068e-05,
+      "loss": 0.167,
+      "step": 6702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.471720818291215e-05,
+      "loss": 0.0917,
+      "step": 6704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.471257983893363e-05,
+      "loss": 4.3045,
+      "step": 6706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.47079514949551e-05,
+      "loss": 0.5585,
+      "step": 6708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.470332315097658e-05,
+      "loss": 1.9404,
+      "step": 6710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4698694806998054e-05,
+      "loss": 0.7362,
+      "step": 6712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.469406646301953e-05,
+      "loss": 4.295,
+      "step": 6714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4689438119041005e-05,
+      "loss": 7.2574,
+      "step": 6716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4684809775062484e-05,
+      "loss": 8.0232,
+      "step": 6718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4680181431083956e-05,
+      "loss": 2.6064,
+      "step": 6720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4675553087105435e-05,
+      "loss": 0.8444,
+      "step": 6722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.467092474312691e-05,
+      "loss": 1.2242,
+      "step": 6724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4666296399148386e-05,
+      "loss": 1.6801,
+      "step": 6726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.466166805516986e-05,
+      "loss": 0.8717,
+      "step": 6728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4657039711191336e-05,
+      "loss": 2.9909,
+      "step": 6730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.465241136721281e-05,
+      "loss": 0.1373,
+      "step": 6732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.464778302323429e-05,
+      "loss": 1.8516,
+      "step": 6734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.464315467925576e-05,
+      "loss": 2.8465,
+      "step": 6736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.463852633527724e-05,
+      "loss": 1.2622,
+      "step": 6738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.463389799129872e-05,
+      "loss": 2.2624,
+      "step": 6740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.462926964732019e-05,
+      "loss": 4.2826,
+      "step": 6742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.462464130334167e-05,
+      "loss": 5.0696,
+      "step": 6744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.462001295936314e-05,
+      "loss": 3.732,
+      "step": 6746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.461538461538462e-05,
+      "loss": 3.1029,
+      "step": 6748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.461075627140609e-05,
+      "loss": 0.7313,
+      "step": 6750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.460612792742757e-05,
+      "loss": 4.4085,
+      "step": 6752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.460149958344904e-05,
+      "loss": 1.8729,
+      "step": 6754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.459687123947052e-05,
+      "loss": 0.0268,
+      "step": 6756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.459224289549199e-05,
+      "loss": 8.3039,
+      "step": 6758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.458761455151347e-05,
+      "loss": 0.2816,
+      "step": 6760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.458298620753494e-05,
+      "loss": 3.4315,
+      "step": 6762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.457835786355642e-05,
+      "loss": 2.2628,
+      "step": 6764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4573729519577894e-05,
+      "loss": 2.7035,
+      "step": 6766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.456910117559937e-05,
+      "loss": 0.0714,
+      "step": 6768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4564472831620845e-05,
+      "loss": 1.8976,
+      "step": 6770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4559844487642324e-05,
+      "loss": 4.1117,
+      "step": 6772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4555216143663796e-05,
+      "loss": 2.436,
+      "step": 6774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4550587799685275e-05,
+      "loss": 3.1147,
+      "step": 6776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.454595945570675e-05,
+      "loss": 2.5759,
+      "step": 6778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4541331111728225e-05,
+      "loss": 0.0236,
+      "step": 6780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4536702767749704e-05,
+      "loss": 0.7883,
+      "step": 6782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4532074423771176e-05,
+      "loss": 0.4863,
+      "step": 6784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4527446079792655e-05,
+      "loss": 2.1716,
+      "step": 6786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.452281773581413e-05,
+      "loss": 0.0029,
+      "step": 6788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4518189391835606e-05,
+      "loss": 2.7143,
+      "step": 6790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.451356104785708e-05,
+      "loss": 0.1565,
+      "step": 6792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.450893270387856e-05,
+      "loss": 4.1106,
+      "step": 6794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.450430435990003e-05,
+      "loss": 0.0971,
+      "step": 6796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.449967601592151e-05,
+      "loss": 4.7502,
+      "step": 6798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.449504767194298e-05,
+      "loss": 0.6609,
+      "step": 6800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.449041932796446e-05,
+      "loss": 1.0782,
+      "step": 6802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.448579098398593e-05,
+      "loss": 0.4513,
+      "step": 6804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.448116264000741e-05,
+      "loss": 2.3011,
+      "step": 6806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.447653429602888e-05,
+      "loss": 2.5149,
+      "step": 6808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.447190595205036e-05,
+      "loss": 3.5451,
+      "step": 6810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.446727760807183e-05,
+      "loss": 0.7672,
+      "step": 6812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.446264926409331e-05,
+      "loss": 3.0574,
+      "step": 6814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.445802092011478e-05,
+      "loss": 0.0037,
+      "step": 6816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.445339257613626e-05,
+      "loss": 1.3192,
+      "step": 6818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.444876423215774e-05,
+      "loss": 3.0898,
+      "step": 6820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.444413588817921e-05,
+      "loss": 7.7,
+      "step": 6822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.443950754420069e-05,
+      "loss": 0.3386,
+      "step": 6824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4434879200222164e-05,
+      "loss": 0.0607,
+      "step": 6826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.443025085624364e-05,
+      "loss": 0.1084,
+      "step": 6828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4425622512265114e-05,
+      "loss": 0.6104,
+      "step": 6830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.442099416828659e-05,
+      "loss": 5.7152,
+      "step": 6832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4416365824308065e-05,
+      "loss": 5.336,
+      "step": 6834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4411737480329544e-05,
+      "loss": 4.6637,
+      "step": 6836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4407109136351016e-05,
+      "loss": 10.8184,
+      "step": 6838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4402480792372495e-05,
+      "loss": 0.3813,
+      "step": 6840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.439785244839397e-05,
+      "loss": 0.9177,
+      "step": 6842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.439322410441544e-05,
+      "loss": 2.3462,
+      "step": 6844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.438859576043692e-05,
+      "loss": 2.2589,
+      "step": 6846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.438396741645839e-05,
+      "loss": 5.7854,
+      "step": 6848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.437933907247987e-05,
+      "loss": 1.9808,
+      "step": 6850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.437471072850134e-05,
+      "loss": 1.918,
+      "step": 6852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.437008238452282e-05,
+      "loss": 2.5519,
+      "step": 6854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.436545404054429e-05,
+      "loss": 4.5179,
+      "step": 6856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.436082569656577e-05,
+      "loss": 1.5449,
+      "step": 6858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.435619735258724e-05,
+      "loss": 0.3575,
+      "step": 6860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.435156900860872e-05,
+      "loss": 1.344,
+      "step": 6862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.434694066463019e-05,
+      "loss": 7.4997,
+      "step": 6864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.434231232065167e-05,
+      "loss": 0.131,
+      "step": 6866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4337683976673144e-05,
+      "loss": 0.3638,
+      "step": 6868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.433305563269462e-05,
+      "loss": 1.4002,
+      "step": 6870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4328427288716095e-05,
+      "loss": 1.0868,
+      "step": 6872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4323798944737574e-05,
+      "loss": 1.301,
+      "step": 6874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4319170600759046e-05,
+      "loss": 0.2701,
+      "step": 6876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4314542256780525e-05,
+      "loss": 0.2271,
+      "step": 6878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4309913912802e-05,
+      "loss": 1.3908,
+      "step": 6880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4305285568823475e-05,
+      "loss": 3.7848,
+      "step": 6882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4300657224844954e-05,
+      "loss": 1.8846,
+      "step": 6884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4296028880866426e-05,
+      "loss": 0.2371,
+      "step": 6886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4291400536887905e-05,
+      "loss": 2.9735,
+      "step": 6888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.428677219290938e-05,
+      "loss": 9.3522,
+      "step": 6890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4282143848930856e-05,
+      "loss": 0.0021,
+      "step": 6892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.427751550495233e-05,
+      "loss": 5.1614,
+      "step": 6894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.427288716097381e-05,
+      "loss": 0.3718,
+      "step": 6896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.426825881699528e-05,
+      "loss": 1.7488,
+      "step": 6898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.426363047301676e-05,
+      "loss": 7.3218,
+      "step": 6900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.425900212903823e-05,
+      "loss": 0.0037,
+      "step": 6902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.425437378505971e-05,
+      "loss": 5.9834,
+      "step": 6904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.424974544108118e-05,
+      "loss": 4.5794,
+      "step": 6906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.424511709710266e-05,
+      "loss": 3.9434,
+      "step": 6908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.424048875312413e-05,
+      "loss": 0.0166,
+      "step": 6910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.423586040914561e-05,
+      "loss": 4.1739,
+      "step": 6912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.423123206516708e-05,
+      "loss": 3.3696,
+      "step": 6914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.422660372118856e-05,
+      "loss": 7.7404,
+      "step": 6916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.422197537721003e-05,
+      "loss": 1.3232,
+      "step": 6918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.421734703323151e-05,
+      "loss": 0.7196,
+      "step": 6920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4212718689252984e-05,
+      "loss": 0.3499,
+      "step": 6922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.420809034527446e-05,
+      "loss": 0.2505,
+      "step": 6924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.420346200129594e-05,
+      "loss": 1.8428,
+      "step": 6926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4198833657317414e-05,
+      "loss": 0.0047,
+      "step": 6928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.419420531333889e-05,
+      "loss": 1.9403,
+      "step": 6930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4189576969360364e-05,
+      "loss": 1.2165,
+      "step": 6932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.418494862538184e-05,
+      "loss": 2.9158,
+      "step": 6934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4180320281403315e-05,
+      "loss": 3.0777,
+      "step": 6936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4175691937424794e-05,
+      "loss": 0.0241,
+      "step": 6938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4171063593446266e-05,
+      "loss": 0.5876,
+      "step": 6940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4166435249467745e-05,
+      "loss": 1.0737,
+      "step": 6942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.416180690548922e-05,
+      "loss": 6.4565,
+      "step": 6944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4157178561510696e-05,
+      "loss": 0.7485,
+      "step": 6946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.415255021753217e-05,
+      "loss": 0.048,
+      "step": 6948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.414792187355365e-05,
+      "loss": 4.263,
+      "step": 6950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.414329352957512e-05,
+      "loss": 1.899,
+      "step": 6952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.41386651855966e-05,
+      "loss": 0.0323,
+      "step": 6954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.413403684161807e-05,
+      "loss": 0.2864,
+      "step": 6956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.412940849763955e-05,
+      "loss": 2.1373,
+      "step": 6958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.412478015366102e-05,
+      "loss": 4.2955,
+      "step": 6960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.41201518096825e-05,
+      "loss": 1.1327,
+      "step": 6962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.411552346570397e-05,
+      "loss": 9.0515,
+      "step": 6964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.411089512172545e-05,
+      "loss": 0.4004,
+      "step": 6966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.410626677774693e-05,
+      "loss": 2.7727,
+      "step": 6968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.41016384337684e-05,
+      "loss": 1.2299,
+      "step": 6970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.409701008978988e-05,
+      "loss": 5.0696,
+      "step": 6972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.409238174581135e-05,
+      "loss": 5.5039,
+      "step": 6974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.408775340183283e-05,
+      "loss": 3.7825,
+      "step": 6976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.40831250578543e-05,
+      "loss": 5.3389,
+      "step": 6978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.407849671387578e-05,
+      "loss": 0.8804,
+      "step": 6980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4073868369897253e-05,
+      "loss": 1.1919,
+      "step": 6982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.406924002591873e-05,
+      "loss": 1.9333,
+      "step": 6984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4064611681940204e-05,
+      "loss": 3.1352,
+      "step": 6986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4059983337961676e-05,
+      "loss": 0.7814,
+      "step": 6988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4055354993983155e-05,
+      "loss": 1.4546,
+      "step": 6990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.405072665000463e-05,
+      "loss": 2.9686,
+      "step": 6992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.4046098306026106e-05,
+      "loss": 0.3886,
+      "step": 6994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.404146996204758e-05,
+      "loss": 3.6217,
+      "step": 6996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.403684161806906e-05,
+      "loss": 1.413,
+      "step": 6998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.403221327409053e-05,
+      "loss": 3.1751,
+      "step": 7000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.402758493011201e-05,
+      "loss": 1.1151,
+      "step": 7002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.402295658613348e-05,
+      "loss": 0.9979,
+      "step": 7004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.401832824215496e-05,
+      "loss": 0.2063,
+      "step": 7006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.401369989817643e-05,
+      "loss": 0.0162,
+      "step": 7008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.400907155419791e-05,
+      "loss": 2.2038,
+      "step": 7010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.400444321021938e-05,
+      "loss": 2.7213,
+      "step": 7012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.399981486624086e-05,
+      "loss": 1.9044,
+      "step": 7014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.399518652226233e-05,
+      "loss": 0.0295,
+      "step": 7016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.399055817828381e-05,
+      "loss": 2.0632,
+      "step": 7018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.398592983430528e-05,
+      "loss": 6.0272,
+      "step": 7020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.398130149032676e-05,
+      "loss": 1.8069,
+      "step": 7022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3976673146348234e-05,
+      "loss": 1.9398,
+      "step": 7024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.397204480236971e-05,
+      "loss": 0.9355,
+      "step": 7026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3967416458391185e-05,
+      "loss": 2.5385,
+      "step": 7028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3962788114412664e-05,
+      "loss": 1.2055,
+      "step": 7030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.395815977043414e-05,
+      "loss": 2.6746,
+      "step": 7032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3953531426455614e-05,
+      "loss": 4.3458,
+      "step": 7034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.394890308247709e-05,
+      "loss": 0.3677,
+      "step": 7036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3944274738498565e-05,
+      "loss": 1.4507,
+      "step": 7038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3939646394520044e-05,
+      "loss": 1.7989,
+      "step": 7040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3935018050541516e-05,
+      "loss": 1.574,
+      "step": 7042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3930389706562995e-05,
+      "loss": 0.2546,
+      "step": 7044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.392576136258447e-05,
+      "loss": 0.9385,
+      "step": 7046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3921133018605946e-05,
+      "loss": 0.0337,
+      "step": 7048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.391650467462742e-05,
+      "loss": 2.7575,
+      "step": 7050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.39118763306489e-05,
+      "loss": 1.3286,
+      "step": 7052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.390724798667037e-05,
+      "loss": 1.6902,
+      "step": 7054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.390261964269185e-05,
+      "loss": 1.3571,
+      "step": 7056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.389799129871332e-05,
+      "loss": 1.3362,
+      "step": 7058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.38933629547348e-05,
+      "loss": 6.8136,
+      "step": 7060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.388873461075627e-05,
+      "loss": 0.602,
+      "step": 7062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.388410626677775e-05,
+      "loss": 0.4026,
+      "step": 7064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.387947792279922e-05,
+      "loss": 2.9295,
+      "step": 7066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.38748495788207e-05,
+      "loss": 2.6847,
+      "step": 7068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.387022123484217e-05,
+      "loss": 5.6971,
+      "step": 7070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.386559289086365e-05,
+      "loss": 1.2851,
+      "step": 7072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.386096454688513e-05,
+      "loss": 0.3928,
+      "step": 7074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.38563362029066e-05,
+      "loss": 4.5452,
+      "step": 7076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.385170785892808e-05,
+      "loss": 5.7598,
+      "step": 7078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.384707951494955e-05,
+      "loss": 0.0005,
+      "step": 7080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.384245117097103e-05,
+      "loss": 4.1093,
+      "step": 7082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3837822826992503e-05,
+      "loss": 2.6514,
+      "step": 7084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.383319448301398e-05,
+      "loss": 0.7469,
+      "step": 7086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3828566139035454e-05,
+      "loss": 2.3298,
+      "step": 7088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.382393779505693e-05,
+      "loss": 1.93,
+      "step": 7090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3819309451078405e-05,
+      "loss": 2.2738,
+      "step": 7092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3814681107099884e-05,
+      "loss": 4.8404,
+      "step": 7094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3810052763121356e-05,
+      "loss": 0.1941,
+      "step": 7096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3805424419142835e-05,
+      "loss": 2.5031,
+      "step": 7098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.380079607516431e-05,
+      "loss": 3.8618,
+      "step": 7100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3796167731185786e-05,
+      "loss": 1.3242,
+      "step": 7102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.379153938720726e-05,
+      "loss": 2.3361,
+      "step": 7104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3786911043228737e-05,
+      "loss": 2.813,
+      "step": 7106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.378228269925021e-05,
+      "loss": 3.3334,
+      "step": 7108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.377765435527169e-05,
+      "loss": 0.9212,
+      "step": 7110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3773026011293166e-05,
+      "loss": 0.2723,
+      "step": 7112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.376839766731464e-05,
+      "loss": 1.5667,
+      "step": 7114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.376376932333612e-05,
+      "loss": 0.1752,
+      "step": 7116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.375914097935759e-05,
+      "loss": 4.3975,
+      "step": 7118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.375451263537907e-05,
+      "loss": 1.962,
+      "step": 7120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.374988429140054e-05,
+      "loss": 1.7204,
+      "step": 7122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.374525594742202e-05,
+      "loss": 0.9031,
+      "step": 7124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.374062760344349e-05,
+      "loss": 1.411,
+      "step": 7126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.373599925946497e-05,
+      "loss": 3.4517,
+      "step": 7128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.373137091548644e-05,
+      "loss": 0.5046,
+      "step": 7130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.372674257150792e-05,
+      "loss": 1.2015,
+      "step": 7132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3722114227529386e-05,
+      "loss": 0.4659,
+      "step": 7134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3717485883550865e-05,
+      "loss": 3.9114,
+      "step": 7136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.371285753957234e-05,
+      "loss": 0.2037,
+      "step": 7138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3708229195593815e-05,
+      "loss": 4.9396,
+      "step": 7140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3703600851615294e-05,
+      "loss": 3.0566,
+      "step": 7142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3698972507636766e-05,
+      "loss": 0.0223,
+      "step": 7144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3694344163658245e-05,
+      "loss": 4.8946,
+      "step": 7146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.368971581967972e-05,
+      "loss": 3.2934,
+      "step": 7148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3685087475701196e-05,
+      "loss": 0.902,
+      "step": 7150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.368045913172267e-05,
+      "loss": 1.2216,
+      "step": 7152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.367583078774415e-05,
+      "loss": 1.2559,
+      "step": 7154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.367120244376562e-05,
+      "loss": 0.7675,
+      "step": 7156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.36665740997871e-05,
+      "loss": 1.1895,
+      "step": 7158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.366194575580857e-05,
+      "loss": 2.4344,
+      "step": 7160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.365731741183005e-05,
+      "loss": 0.6809,
+      "step": 7162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.365268906785152e-05,
+      "loss": 1.7925,
+      "step": 7164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3648060723873e-05,
+      "loss": 3.1567,
+      "step": 7166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.364343237989447e-05,
+      "loss": 1.4077,
+      "step": 7168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.363880403591595e-05,
+      "loss": 0.1837,
+      "step": 7170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.363417569193742e-05,
+      "loss": 2.6869,
+      "step": 7172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.36295473479589e-05,
+      "loss": 0.537,
+      "step": 7174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.362491900398037e-05,
+      "loss": 2.3665,
+      "step": 7176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.362029066000185e-05,
+      "loss": 2.6044,
+      "step": 7178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.361566231602333e-05,
+      "loss": 3.6506,
+      "step": 7180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.36110339720448e-05,
+      "loss": 4.11,
+      "step": 7182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.360640562806628e-05,
+      "loss": 4.7732,
+      "step": 7184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3601777284087754e-05,
+      "loss": 0.306,
+      "step": 7186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.359714894010923e-05,
+      "loss": 7.5727,
+      "step": 7188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3592520596130704e-05,
+      "loss": 0.7458,
+      "step": 7190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.358789225215218e-05,
+      "loss": 0.0066,
+      "step": 7192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3583263908173655e-05,
+      "loss": 2.8496,
+      "step": 7194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3578635564195134e-05,
+      "loss": 1.8856,
+      "step": 7196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3574007220216606e-05,
+      "loss": 1.1265,
+      "step": 7198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3569378876238085e-05,
+      "loss": 3.5462,
+      "step": 7200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.356475053225956e-05,
+      "loss": 1.3608,
+      "step": 7202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3560122188281036e-05,
+      "loss": 0.9876,
+      "step": 7204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.355549384430251e-05,
+      "loss": 5.7284,
+      "step": 7206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3550865500323987e-05,
+      "loss": 2.0203,
+      "step": 7208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.354623715634546e-05,
+      "loss": 0.4722,
+      "step": 7210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.354160881236694e-05,
+      "loss": 5.8665,
+      "step": 7212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.353698046838841e-05,
+      "loss": 6.9109,
+      "step": 7214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.353235212440989e-05,
+      "loss": 2.6135,
+      "step": 7216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.352772378043137e-05,
+      "loss": 1.4161,
+      "step": 7218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.352309543645284e-05,
+      "loss": 5.1989,
+      "step": 7220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.351846709247432e-05,
+      "loss": 3.1732,
+      "step": 7222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.351383874849579e-05,
+      "loss": 2.1165,
+      "step": 7224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.350921040451727e-05,
+      "loss": 2.8775,
+      "step": 7226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.350458206053874e-05,
+      "loss": 0.3382,
+      "step": 7228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.349995371656022e-05,
+      "loss": 1.7314,
+      "step": 7230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.349532537258169e-05,
+      "loss": 0.6708,
+      "step": 7232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.349069702860317e-05,
+      "loss": 3.3461,
+      "step": 7234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.348606868462464e-05,
+      "loss": 2.8509,
+      "step": 7236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.348144034064612e-05,
+      "loss": 0.0143,
+      "step": 7238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.347681199666759e-05,
+      "loss": 0.0074,
+      "step": 7240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.347218365268907e-05,
+      "loss": 3.2902,
+      "step": 7242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3467555308710544e-05,
+      "loss": 4.074,
+      "step": 7244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.346292696473202e-05,
+      "loss": 0.3031,
+      "step": 7246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3458298620753495e-05,
+      "loss": 0.0003,
+      "step": 7248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3453670276774974e-05,
+      "loss": 7.7183,
+      "step": 7250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3449041932796446e-05,
+      "loss": 0.1247,
+      "step": 7252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3444413588817925e-05,
+      "loss": 2.4372,
+      "step": 7254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.34397852448394e-05,
+      "loss": 4.2571,
+      "step": 7256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3435156900860876e-05,
+      "loss": 1.3378,
+      "step": 7258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3430528556882354e-05,
+      "loss": 2.3891,
+      "step": 7260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3425900212903826e-05,
+      "loss": 3.7999,
+      "step": 7262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3421271868925305e-05,
+      "loss": 4.2878,
+      "step": 7264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.341664352494678e-05,
+      "loss": 1.1382,
+      "step": 7266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3412015180968256e-05,
+      "loss": 0.9325,
+      "step": 7268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.340738683698973e-05,
+      "loss": 3.8178,
+      "step": 7270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.340275849301121e-05,
+      "loss": 1.6316,
+      "step": 7272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.339813014903268e-05,
+      "loss": 3.4476,
+      "step": 7274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.339350180505416e-05,
+      "loss": 1.7407,
+      "step": 7276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.338887346107562e-05,
+      "loss": 0.6441,
+      "step": 7278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.33842451170971e-05,
+      "loss": 2.0852,
+      "step": 7280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3379616773118574e-05,
+      "loss": 0.0673,
+      "step": 7282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.337498842914005e-05,
+      "loss": 4.3759,
+      "step": 7284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.337036008516153e-05,
+      "loss": 4.2592,
+      "step": 7286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3365731741183004e-05,
+      "loss": 0.0005,
+      "step": 7288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.336110339720448e-05,
+      "loss": 1.1348,
+      "step": 7290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3356475053225954e-05,
+      "loss": 3.2019,
+      "step": 7292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.335184670924743e-05,
+      "loss": 2.9673,
+      "step": 7294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3347218365268905e-05,
+      "loss": 1.0847,
+      "step": 7296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3342590021290384e-05,
+      "loss": 0.0002,
+      "step": 7298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3337961677311856e-05,
+      "loss": 0.9707,
+      "step": 7300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.2439,
+      "step": 7302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.332870498935481e-05,
+      "loss": 5.849,
+      "step": 7304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3324076645376286e-05,
+      "loss": 4.1112,
+      "step": 7306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.331944830139776e-05,
+      "loss": 2.6247,
+      "step": 7308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.331481995741924e-05,
+      "loss": 3.208,
+      "step": 7310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.331019161344071e-05,
+      "loss": 1.191,
+      "step": 7312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.330556326946219e-05,
+      "loss": 2.5727,
+      "step": 7314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.330093492548366e-05,
+      "loss": 1.1109,
+      "step": 7316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.329630658150514e-05,
+      "loss": 0.001,
+      "step": 7318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.329167823752661e-05,
+      "loss": 4.876,
+      "step": 7320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.328704989354809e-05,
+      "loss": 2.1978,
+      "step": 7322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.328242154956956e-05,
+      "loss": 0.0031,
+      "step": 7324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.327779320559104e-05,
+      "loss": 1.7872,
+      "step": 7326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.327316486161252e-05,
+      "loss": 2.6104,
+      "step": 7328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.326853651763399e-05,
+      "loss": 2.3763,
+      "step": 7330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.326390817365547e-05,
+      "loss": 1.7005,
+      "step": 7332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.325927982967694e-05,
+      "loss": 2.6769,
+      "step": 7334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.325465148569842e-05,
+      "loss": 2.636,
+      "step": 7336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.325002314171989e-05,
+      "loss": 4.6505,
+      "step": 7338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.324539479774137e-05,
+      "loss": 1.8923,
+      "step": 7340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3240766453762843e-05,
+      "loss": 3.2733,
+      "step": 7342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.323613810978432e-05,
+      "loss": 1.8884,
+      "step": 7344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3231509765805794e-05,
+      "loss": 2.9984,
+      "step": 7346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.322688142182727e-05,
+      "loss": 0.0067,
+      "step": 7348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3222253077848745e-05,
+      "loss": 2.7365,
+      "step": 7350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3217624733870224e-05,
+      "loss": 0.9992,
+      "step": 7352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3212996389891696e-05,
+      "loss": 0.3307,
+      "step": 7354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3208368045913175e-05,
+      "loss": 5.305,
+      "step": 7356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.320373970193465e-05,
+      "loss": 3.4544,
+      "step": 7358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3199111357956126e-05,
+      "loss": 1.553,
+      "step": 7360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.31944830139776e-05,
+      "loss": 1.5212,
+      "step": 7362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3189854669999076e-05,
+      "loss": 6.3907,
+      "step": 7364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3185226326020555e-05,
+      "loss": 0.9458,
+      "step": 7366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.318059798204203e-05,
+      "loss": 0.6087,
+      "step": 7368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3175969638063506e-05,
+      "loss": 0.6988,
+      "step": 7370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.317134129408498e-05,
+      "loss": 3.1646,
+      "step": 7372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.316671295010646e-05,
+      "loss": 1.7048,
+      "step": 7374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.316208460612793e-05,
+      "loss": 4.8524,
+      "step": 7376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.315745626214941e-05,
+      "loss": 0.1489,
+      "step": 7378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.315282791817088e-05,
+      "loss": 5.9255,
+      "step": 7380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.314819957419236e-05,
+      "loss": 2.744,
+      "step": 7382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.314357123021383e-05,
+      "loss": 0.8175,
+      "step": 7384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.313894288623531e-05,
+      "loss": 4.4632,
+      "step": 7386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.313431454225678e-05,
+      "loss": 0.0195,
+      "step": 7388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.312968619827826e-05,
+      "loss": 4.3856,
+      "step": 7390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.312505785429973e-05,
+      "loss": 2.9194,
+      "step": 7392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.312042951032121e-05,
+      "loss": 2.8344,
+      "step": 7394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.311580116634268e-05,
+      "loss": 1.4813,
+      "step": 7396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.311117282236416e-05,
+      "loss": 5.513,
+      "step": 7398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3106544478385634e-05,
+      "loss": 1.9541,
+      "step": 7400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.310191613440711e-05,
+      "loss": 1.632,
+      "step": 7402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3097287790428585e-05,
+      "loss": 1.4746,
+      "step": 7404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3092659446450064e-05,
+      "loss": 1.1526,
+      "step": 7406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.308803110247154e-05,
+      "loss": 1.2376,
+      "step": 7408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3083402758493015e-05,
+      "loss": 1.5268,
+      "step": 7410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3078774414514493e-05,
+      "loss": 3.5297,
+      "step": 7412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3074146070535965e-05,
+      "loss": 4.1224,
+      "step": 7414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3069517726557444e-05,
+      "loss": 0.3747,
+      "step": 7416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3064889382578916e-05,
+      "loss": 0.001,
+      "step": 7418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3060261038600395e-05,
+      "loss": 0.5452,
+      "step": 7420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.305563269462187e-05,
+      "loss": 0.515,
+      "step": 7422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.305100435064334e-05,
+      "loss": 0.3612,
+      "step": 7424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.304637600666481e-05,
+      "loss": 0.4824,
+      "step": 7426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.304174766268629e-05,
+      "loss": 2.3856,
+      "step": 7428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.303711931870776e-05,
+      "loss": 2.0039,
+      "step": 7430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.303249097472924e-05,
+      "loss": 2.5919,
+      "step": 7432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.302786263075072e-05,
+      "loss": 3.8647,
+      "step": 7434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.302323428677219e-05,
+      "loss": 5.0429,
+      "step": 7436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.301860594279367e-05,
+      "loss": 1.8329,
+      "step": 7438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.301397759881514e-05,
+      "loss": 5.4466,
+      "step": 7440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.300934925483662e-05,
+      "loss": 1.1203,
+      "step": 7442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3004720910858093e-05,
+      "loss": 0.2703,
+      "step": 7444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.300009256687957e-05,
+      "loss": 0.3675,
+      "step": 7446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2995464222901044e-05,
+      "loss": 1.6426,
+      "step": 7448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.299083587892252e-05,
+      "loss": 2.4336,
+      "step": 7450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2986207534943995e-05,
+      "loss": 0.0036,
+      "step": 7452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2981579190965474e-05,
+      "loss": 5.8081,
+      "step": 7454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2976950846986946e-05,
+      "loss": 1.5859,
+      "step": 7456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2972322503008425e-05,
+      "loss": 4.5346,
+      "step": 7458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.29676941590299e-05,
+      "loss": 2.6264,
+      "step": 7460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2963065815051376e-05,
+      "loss": 1.763,
+      "step": 7462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.295843747107285e-05,
+      "loss": 1.2659,
+      "step": 7464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2953809127094327e-05,
+      "loss": 3.7921,
+      "step": 7466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.29491807831158e-05,
+      "loss": 1.4024,
+      "step": 7468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.294455243913728e-05,
+      "loss": 3.1618,
+      "step": 7470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2939924095158756e-05,
+      "loss": 0.4153,
+      "step": 7472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.293529575118023e-05,
+      "loss": 2.1999,
+      "step": 7474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.293066740720171e-05,
+      "loss": 2.9299,
+      "step": 7476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.292603906322318e-05,
+      "loss": 1.7345,
+      "step": 7478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.292141071924466e-05,
+      "loss": 0.0303,
+      "step": 7480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.291678237526613e-05,
+      "loss": 1.5046,
+      "step": 7482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.291215403128761e-05,
+      "loss": 0.1274,
+      "step": 7484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.290752568730908e-05,
+      "loss": 3.2658,
+      "step": 7486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.290289734333056e-05,
+      "loss": 0.0738,
+      "step": 7488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.289826899935203e-05,
+      "loss": 4.0224,
+      "step": 7490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.289364065537351e-05,
+      "loss": 4.4324,
+      "step": 7492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.288901231139498e-05,
+      "loss": 0.0003,
+      "step": 7494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.288438396741646e-05,
+      "loss": 3.4167,
+      "step": 7496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.287975562343793e-05,
+      "loss": 0.9708,
+      "step": 7498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.287512727945941e-05,
+      "loss": 2.3268,
+      "step": 7500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2870498935480884e-05,
+      "loss": 0.0002,
+      "step": 7502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.286587059150236e-05,
+      "loss": 0.0009,
+      "step": 7504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2861242247523835e-05,
+      "loss": 0.9279,
+      "step": 7506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2856613903545314e-05,
+      "loss": 5.5095,
+      "step": 7508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2851985559566786e-05,
+      "loss": 0.5624,
+      "step": 7510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2847357215588265e-05,
+      "loss": 2.5446,
+      "step": 7512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2842728871609743e-05,
+      "loss": 1.4876,
+      "step": 7514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2838100527631216e-05,
+      "loss": 1.4801,
+      "step": 7516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2833472183652694e-05,
+      "loss": 5.8203,
+      "step": 7518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2828843839674166e-05,
+      "loss": 6.354,
+      "step": 7520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2824215495695645e-05,
+      "loss": 7.8923,
+      "step": 7522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.281958715171712e-05,
+      "loss": 3.2861,
+      "step": 7524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2814958807738596e-05,
+      "loss": 5.3633,
+      "step": 7526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.281033046376007e-05,
+      "loss": 1.9322,
+      "step": 7528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.280570211978155e-05,
+      "loss": 1.6802,
+      "step": 7530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.280107377580302e-05,
+      "loss": 0.0897,
+      "step": 7532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.27964454318245e-05,
+      "loss": 0.7123,
+      "step": 7534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.279181708784597e-05,
+      "loss": 3.0438,
+      "step": 7536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.278718874386745e-05,
+      "loss": 2.4621,
+      "step": 7538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.278256039988892e-05,
+      "loss": 2.2974,
+      "step": 7540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.27779320559104e-05,
+      "loss": 2.7314,
+      "step": 7542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.277330371193187e-05,
+      "loss": 2.5775,
+      "step": 7544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.276867536795335e-05,
+      "loss": 1.257,
+      "step": 7546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.276404702397482e-05,
+      "loss": 1.7973,
+      "step": 7548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.27594186799963e-05,
+      "loss": 1.2464,
+      "step": 7550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.275479033601778e-05,
+      "loss": 0.8818,
+      "step": 7552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.275016199203925e-05,
+      "loss": 2.5286,
+      "step": 7554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.274553364806073e-05,
+      "loss": 3.3742,
+      "step": 7556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.27409053040822e-05,
+      "loss": 7.9011,
+      "step": 7558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.273627696010368e-05,
+      "loss": 0.0036,
+      "step": 7560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2731648616125154e-05,
+      "loss": 1.2091,
+      "step": 7562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.272702027214663e-05,
+      "loss": 3.9492,
+      "step": 7564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2722391928168105e-05,
+      "loss": 1.1728,
+      "step": 7566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2717763584189577e-05,
+      "loss": 1.2549,
+      "step": 7568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.271313524021105e-05,
+      "loss": 2.2042,
+      "step": 7570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.270850689623253e-05,
+      "loss": 1.6174,
+      "step": 7572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2703878552254e-05,
+      "loss": 6.3657,
+      "step": 7574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.269925020827548e-05,
+      "loss": 6.7479,
+      "step": 7576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.269462186429696e-05,
+      "loss": 3.0452,
+      "step": 7578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.268999352031843e-05,
+      "loss": 0.9804,
+      "step": 7580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.268536517633991e-05,
+      "loss": 0.1887,
+      "step": 7582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.268073683236138e-05,
+      "loss": 0.4592,
+      "step": 7584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.267610848838286e-05,
+      "loss": 4.3856,
+      "step": 7586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.267148014440433e-05,
+      "loss": 0.8874,
+      "step": 7588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.266685180042581e-05,
+      "loss": 6.2125,
+      "step": 7590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.266222345644728e-05,
+      "loss": 2.9394,
+      "step": 7592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.265759511246876e-05,
+      "loss": 2.6411,
+      "step": 7594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.265296676849023e-05,
+      "loss": 2.8189,
+      "step": 7596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.264833842451171e-05,
+      "loss": 2.3853,
+      "step": 7598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.264371008053318e-05,
+      "loss": 2.3411,
+      "step": 7600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.263908173655466e-05,
+      "loss": 2.846,
+      "step": 7602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2634453392576134e-05,
+      "loss": 2.1812,
+      "step": 7604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.262982504859761e-05,
+      "loss": 3.8997,
+      "step": 7606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2625196704619085e-05,
+      "loss": 6.9646,
+      "step": 7608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2620568360640564e-05,
+      "loss": 6.6754,
+      "step": 7610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2615940016662036e-05,
+      "loss": 6.8442,
+      "step": 7612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2611311672683515e-05,
+      "loss": 0.9615,
+      "step": 7614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.260668332870499e-05,
+      "loss": 1.3182,
+      "step": 7616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2602054984726466e-05,
+      "loss": 0.1915,
+      "step": 7618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2597426640747944e-05,
+      "loss": 0.0016,
+      "step": 7620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2592798296769416e-05,
+      "loss": 0.049,
+      "step": 7622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2588169952790895e-05,
+      "loss": 0.4228,
+      "step": 7624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.258354160881237e-05,
+      "loss": 0.2418,
+      "step": 7626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2578913264833846e-05,
+      "loss": 2.0256,
+      "step": 7628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.257428492085532e-05,
+      "loss": 3.1484,
+      "step": 7630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.25696565768768e-05,
+      "loss": 8.1448,
+      "step": 7632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.256502823289827e-05,
+      "loss": 0.9573,
+      "step": 7634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.256039988891975e-05,
+      "loss": 3.5413,
+      "step": 7636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.255577154494122e-05,
+      "loss": 3.1948,
+      "step": 7638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.25511432009627e-05,
+      "loss": 0.003,
+      "step": 7640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.254651485698417e-05,
+      "loss": 0.0518,
+      "step": 7642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.254188651300565e-05,
+      "loss": 0.0551,
+      "step": 7644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.253725816902712e-05,
+      "loss": 1.1148,
+      "step": 7646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.25326298250486e-05,
+      "loss": 7.6741,
+      "step": 7648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.252800148107007e-05,
+      "loss": 3.7683,
+      "step": 7650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.252337313709155e-05,
+      "loss": 0.7159,
+      "step": 7652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.251874479311302e-05,
+      "loss": 5.7249,
+      "step": 7654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.25141164491345e-05,
+      "loss": 0.0016,
+      "step": 7656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2509488105155974e-05,
+      "loss": 6.2266,
+      "step": 7658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.250485976117745e-05,
+      "loss": 0.8546,
+      "step": 7660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.250023141719893e-05,
+      "loss": 1.9061,
+      "step": 7662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2495603073220404e-05,
+      "loss": 0.6941,
+      "step": 7664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.249097472924188e-05,
+      "loss": 0.907,
+      "step": 7666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2486346385263355e-05,
+      "loss": 2.7791,
+      "step": 7668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.248171804128483e-05,
+      "loss": 2.7647,
+      "step": 7670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2477089697306305e-05,
+      "loss": 2.8938,
+      "step": 7672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2472461353327784e-05,
+      "loss": 6.2322,
+      "step": 7674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2467833009349256e-05,
+      "loss": 4.5319,
+      "step": 7676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2463204665370735e-05,
+      "loss": 1.3444,
+      "step": 7678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.245857632139221e-05,
+      "loss": 3.1963,
+      "step": 7680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2453947977413686e-05,
+      "loss": 1.0827,
+      "step": 7682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.244931963343516e-05,
+      "loss": 0.4566,
+      "step": 7684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.244469128945664e-05,
+      "loss": 3.8082,
+      "step": 7686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.244006294547811e-05,
+      "loss": 6.3307,
+      "step": 7688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.243543460149959e-05,
+      "loss": 0.8508,
+      "step": 7690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.243080625752106e-05,
+      "loss": 1.761,
+      "step": 7692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.242617791354254e-05,
+      "loss": 1.5045,
+      "step": 7694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.242154956956401e-05,
+      "loss": 1.7304,
+      "step": 7696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.241692122558549e-05,
+      "loss": 4.739,
+      "step": 7698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.241229288160697e-05,
+      "loss": 1.3505,
+      "step": 7700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.240766453762844e-05,
+      "loss": 1.7968,
+      "step": 7702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.240303619364992e-05,
+      "loss": 2.1542,
+      "step": 7704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.239840784967139e-05,
+      "loss": 2.2459,
+      "step": 7706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.239377950569287e-05,
+      "loss": 0.0014,
+      "step": 7708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.238915116171434e-05,
+      "loss": 0.6826,
+      "step": 7710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.238452281773582e-05,
+      "loss": 1.0702,
+      "step": 7712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2379894473757286e-05,
+      "loss": 1.6562,
+      "step": 7714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2375266129778765e-05,
+      "loss": 0.0004,
+      "step": 7716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.237063778580024e-05,
+      "loss": 6.2955,
+      "step": 7718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2366009441821716e-05,
+      "loss": 2.9639,
+      "step": 7720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.236138109784319e-05,
+      "loss": 0.2657,
+      "step": 7722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2356752753864666e-05,
+      "loss": 3.5979,
+      "step": 7724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2352124409886145e-05,
+      "loss": 3.7538,
+      "step": 7726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.234749606590762e-05,
+      "loss": 0.8819,
+      "step": 7728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2342867721929096e-05,
+      "loss": 1.2886,
+      "step": 7730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.233823937795057e-05,
+      "loss": 2.257,
+      "step": 7732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.233361103397205e-05,
+      "loss": 1.3383,
+      "step": 7734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.232898268999352e-05,
+      "loss": 2.126,
+      "step": 7736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2324354346015e-05,
+      "loss": 0.9785,
+      "step": 7738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.231972600203647e-05,
+      "loss": 3.6291,
+      "step": 7740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.231509765805795e-05,
+      "loss": 2.3421,
+      "step": 7742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.231046931407942e-05,
+      "loss": 6.215,
+      "step": 7744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.23058409701009e-05,
+      "loss": 0.2446,
+      "step": 7746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.230121262612237e-05,
+      "loss": 2.4229,
+      "step": 7748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.229658428214385e-05,
+      "loss": 1.3068,
+      "step": 7750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.229195593816532e-05,
+      "loss": 1.149,
+      "step": 7752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.22873275941868e-05,
+      "loss": 2.2274,
+      "step": 7754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.228269925020827e-05,
+      "loss": 1.0459,
+      "step": 7756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.227807090622975e-05,
+      "loss": 1.5083,
+      "step": 7758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2273442562251224e-05,
+      "loss": 3.5236,
+      "step": 7760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.22688142182727e-05,
+      "loss": 4.633,
+      "step": 7762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2264185874294175e-05,
+      "loss": 1.9962,
+      "step": 7764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2259557530315654e-05,
+      "loss": 1.2083,
+      "step": 7766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.225492918633713e-05,
+      "loss": 0.0458,
+      "step": 7768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2250300842358605e-05,
+      "loss": 0.5635,
+      "step": 7770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2245672498380083e-05,
+      "loss": 4.6251,
+      "step": 7772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2241044154401555e-05,
+      "loss": 0.3963,
+      "step": 7774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2236415810423034e-05,
+      "loss": 1.3559,
+      "step": 7776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2231787466444506e-05,
+      "loss": 6.4317,
+      "step": 7778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2227159122465985e-05,
+      "loss": 6.3656,
+      "step": 7780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.222253077848746e-05,
+      "loss": 2.3569,
+      "step": 7782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2217902434508936e-05,
+      "loss": 0.0869,
+      "step": 7784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.221327409053041e-05,
+      "loss": 1.3258,
+      "step": 7786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.220864574655189e-05,
+      "loss": 3.016,
+      "step": 7788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.220401740257336e-05,
+      "loss": 1.9786,
+      "step": 7790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.219938905859484e-05,
+      "loss": 0.078,
+      "step": 7792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.219476071461631e-05,
+      "loss": 3.2488,
+      "step": 7794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.219013237063779e-05,
+      "loss": 3.1158,
+      "step": 7796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.218550402665926e-05,
+      "loss": 0.369,
+      "step": 7798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.218087568268074e-05,
+      "loss": 3.4031,
+      "step": 7800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.217624733870221e-05,
+      "loss": 2.4213,
+      "step": 7802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.217161899472369e-05,
+      "loss": 1.7498,
+      "step": 7804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.216699065074517e-05,
+      "loss": 1.4593,
+      "step": 7806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.216236230676664e-05,
+      "loss": 0.0629,
+      "step": 7808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.215773396278812e-05,
+      "loss": 0.0774,
+      "step": 7810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.215310561880959e-05,
+      "loss": 1.5424,
+      "step": 7812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.214847727483107e-05,
+      "loss": 3.7143,
+      "step": 7814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.214384893085254e-05,
+      "loss": 0.0011,
+      "step": 7816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.213922058687402e-05,
+      "loss": 3.8993,
+      "step": 7818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2134592242895494e-05,
+      "loss": 2.0086,
+      "step": 7820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.212996389891697e-05,
+      "loss": 0.0676,
+      "step": 7822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2125335554938444e-05,
+      "loss": 3.4925,
+      "step": 7824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.212070721095992e-05,
+      "loss": 3.1057,
+      "step": 7826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2116078866981395e-05,
+      "loss": 2.3035,
+      "step": 7828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2111450523002874e-05,
+      "loss": 0.0057,
+      "step": 7830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2106822179024346e-05,
+      "loss": 0.0014,
+      "step": 7832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2102193835045825e-05,
+      "loss": 2.5449,
+      "step": 7834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.20975654910673e-05,
+      "loss": 7.0482,
+      "step": 7836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2092937147088776e-05,
+      "loss": 1.299,
+      "step": 7838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.208830880311025e-05,
+      "loss": 3.0205,
+      "step": 7840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.208368045913173e-05,
+      "loss": 1.9514,
+      "step": 7842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.20790521151532e-05,
+      "loss": 0.0586,
+      "step": 7844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.207442377117468e-05,
+      "loss": 4.9048,
+      "step": 7846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2069795427196156e-05,
+      "loss": 3.5076,
+      "step": 7848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.206516708321763e-05,
+      "loss": 2.3847,
+      "step": 7850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.206053873923911e-05,
+      "loss": 1.0066,
+      "step": 7852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.205591039526058e-05,
+      "loss": 0.3838,
+      "step": 7854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.205128205128206e-05,
+      "loss": 2.1757,
+      "step": 7856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.204665370730352e-05,
+      "loss": 6.9891,
+      "step": 7858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2042025363325e-05,
+      "loss": 5.0924,
+      "step": 7860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2037397019346474e-05,
+      "loss": 2.6594,
+      "step": 7862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.203276867536795e-05,
+      "loss": 4.203,
+      "step": 7864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2028140331389425e-05,
+      "loss": 0.2276,
+      "step": 7866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2023511987410904e-05,
+      "loss": 0.8025,
+      "step": 7868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2018883643432376e-05,
+      "loss": 2.8032,
+      "step": 7870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2014255299453855e-05,
+      "loss": 4.0629,
+      "step": 7872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2009626955475333e-05,
+      "loss": 1.5313,
+      "step": 7874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2004998611496805e-05,
+      "loss": 4.9594,
+      "step": 7876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2000370267518284e-05,
+      "loss": 1.0184,
+      "step": 7878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1995741923539756e-05,
+      "loss": 0.0308,
+      "step": 7880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1991113579561235e-05,
+      "loss": 0.8684,
+      "step": 7882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.198648523558271e-05,
+      "loss": 0.8938,
+      "step": 7884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1981856891604186e-05,
+      "loss": 0.9181,
+      "step": 7886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.197722854762566e-05,
+      "loss": 1.4285,
+      "step": 7888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.197260020364714e-05,
+      "loss": 5.1826,
+      "step": 7890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.196797185966861e-05,
+      "loss": 0.0998,
+      "step": 7892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.196334351569009e-05,
+      "loss": 1.434,
+      "step": 7894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.195871517171156e-05,
+      "loss": 0.3821,
+      "step": 7896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.195408682773304e-05,
+      "loss": 0.0026,
+      "step": 7898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.194945848375451e-05,
+      "loss": 1.3475,
+      "step": 7900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.194483013977599e-05,
+      "loss": 2.1203,
+      "step": 7902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.194020179579746e-05,
+      "loss": 2.4074,
+      "step": 7904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.193557345181894e-05,
+      "loss": 4.6178,
+      "step": 7906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.193094510784041e-05,
+      "loss": 0.5448,
+      "step": 7908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.192631676386189e-05,
+      "loss": 1.4482,
+      "step": 7910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.192168841988337e-05,
+      "loss": 3.7371,
+      "step": 7912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.191706007590484e-05,
+      "loss": 2.1748,
+      "step": 7914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.191243173192632e-05,
+      "loss": 0.9842,
+      "step": 7916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.190780338794779e-05,
+      "loss": 3.1469,
+      "step": 7918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.190317504396927e-05,
+      "loss": 2.5506,
+      "step": 7920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1898546699990744e-05,
+      "loss": 0.2038,
+      "step": 7922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.189391835601222e-05,
+      "loss": 5.0057,
+      "step": 7924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1889290012033695e-05,
+      "loss": 0.0036,
+      "step": 7926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.188466166805517e-05,
+      "loss": 2.5364,
+      "step": 7928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1880033324076645e-05,
+      "loss": 0.0564,
+      "step": 7930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1875404980098124e-05,
+      "loss": 2.7278,
+      "step": 7932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1870776636119596e-05,
+      "loss": 1.5835,
+      "step": 7934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1866148292141075e-05,
+      "loss": 1.9478,
+      "step": 7936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.186151994816255e-05,
+      "loss": 5.7761,
+      "step": 7938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1856891604184026e-05,
+      "loss": 0.8422,
+      "step": 7940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.18522632602055e-05,
+      "loss": 0.3377,
+      "step": 7942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.184763491622698e-05,
+      "loss": 0.9974,
+      "step": 7944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.184300657224845e-05,
+      "loss": 4.1441,
+      "step": 7946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.183837822826993e-05,
+      "loss": 4.9168,
+      "step": 7948
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.18337498842914e-05,
+      "loss": 0.0353,
+      "step": 7950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.182912154031288e-05,
+      "loss": 4.4639,
+      "step": 7952
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.182449319633436e-05,
+      "loss": 2.6699,
+      "step": 7954
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.181986485235583e-05,
+      "loss": 3.1607,
+      "step": 7956
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.181523650837731e-05,
+      "loss": 0.0173,
+      "step": 7958
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.181060816439878e-05,
+      "loss": 4.8739,
+      "step": 7960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.180597982042026e-05,
+      "loss": 3.1121,
+      "step": 7962
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.180135147644173e-05,
+      "loss": 4.6104,
+      "step": 7964
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.179672313246321e-05,
+      "loss": 3.0082,
+      "step": 7966
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.179209478848468e-05,
+      "loss": 4.4288,
+      "step": 7968
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.178746644450616e-05,
+      "loss": 0.5654,
+      "step": 7970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.178283810052763e-05,
+      "loss": 0.0025,
+      "step": 7972
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.177820975654911e-05,
+      "loss": 0.8769,
+      "step": 7974
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1773581412570584e-05,
+      "loss": 0.7051,
+      "step": 7976
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.176895306859206e-05,
+      "loss": 2.7188,
+      "step": 7978
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1764324724613534e-05,
+      "loss": 1.4885,
+      "step": 7980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.175969638063501e-05,
+      "loss": 1.6505,
+      "step": 7982
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1755068036656485e-05,
+      "loss": 4.7467,
+      "step": 7984
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1750439692677964e-05,
+      "loss": 0.0204,
+      "step": 7986
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1745811348699436e-05,
+      "loss": 7.0076,
+      "step": 7988
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1741183004720915e-05,
+      "loss": 2.5548,
+      "step": 7990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.173655466074239e-05,
+      "loss": 2.8503,
+      "step": 7992
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1731926316763866e-05,
+      "loss": 2.3065,
+      "step": 7994
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1727297972785345e-05,
+      "loss": 2.2659,
+      "step": 7996
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1722669628806817e-05,
+      "loss": 2.9816,
+      "step": 7998
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1718041284828295e-05,
+      "loss": 4.486,
+      "step": 8000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.171341294084977e-05,
+      "loss": 1.2234,
+      "step": 8002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.170878459687124e-05,
+      "loss": 0.0257,
+      "step": 8004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.170415625289271e-05,
+      "loss": 1.828,
+      "step": 8006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.169952790891419e-05,
+      "loss": 1.32,
+      "step": 8008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.169489956493566e-05,
+      "loss": 3.9426,
+      "step": 8010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.169027122095714e-05,
+      "loss": 2.2307,
+      "step": 8012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.168564287697861e-05,
+      "loss": 0.0018,
+      "step": 8014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.168101453300009e-05,
+      "loss": 0.0017,
+      "step": 8016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1676386189021564e-05,
+      "loss": 1.6364,
+      "step": 8018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.167175784504304e-05,
+      "loss": 4.2082,
+      "step": 8020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.166712950106452e-05,
+      "loss": 1.4615,
+      "step": 8022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1662501157085994e-05,
+      "loss": 1.9671,
+      "step": 8024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.165787281310747e-05,
+      "loss": 0.0108,
+      "step": 8026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1653244469128945e-05,
+      "loss": 1.3288,
+      "step": 8028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.164861612515042e-05,
+      "loss": 1.6305,
+      "step": 8030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1643987781171895e-05,
+      "loss": 2.206,
+      "step": 8032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1639359437193374e-05,
+      "loss": 0.8777,
+      "step": 8034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1634731093214846e-05,
+      "loss": 4.2899,
+      "step": 8036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1630102749236325e-05,
+      "loss": 4.7501,
+      "step": 8038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.16254744052578e-05,
+      "loss": 1.5673,
+      "step": 8040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1620846061279276e-05,
+      "loss": 6.0869,
+      "step": 8042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.161621771730075e-05,
+      "loss": 0.703,
+      "step": 8044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.161158937332223e-05,
+      "loss": 7.0084,
+      "step": 8046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.16069610293437e-05,
+      "loss": 1.9958,
+      "step": 8048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.160233268536518e-05,
+      "loss": 3.9442,
+      "step": 8050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.159770434138665e-05,
+      "loss": 0.0002,
+      "step": 8052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.159307599740813e-05,
+      "loss": 1.7507,
+      "step": 8054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.15884476534296e-05,
+      "loss": 0.0896,
+      "step": 8056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.158381930945108e-05,
+      "loss": 0.0474,
+      "step": 8058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.157919096547256e-05,
+      "loss": 0.4296,
+      "step": 8060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.157456262149403e-05,
+      "loss": 1.231,
+      "step": 8062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.156993427751551e-05,
+      "loss": 3.2459,
+      "step": 8064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.156530593353698e-05,
+      "loss": 2.3049,
+      "step": 8066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.156067758955846e-05,
+      "loss": 2.2557,
+      "step": 8068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.155604924557993e-05,
+      "loss": 0.0422,
+      "step": 8070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.155142090160141e-05,
+      "loss": 5.9104,
+      "step": 8072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.154679255762288e-05,
+      "loss": 2.7001,
+      "step": 8074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.154216421364436e-05,
+      "loss": 2.5937,
+      "step": 8076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1537535869665834e-05,
+      "loss": 2.3223,
+      "step": 8078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.153290752568731e-05,
+      "loss": 2.2062,
+      "step": 8080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1528279181708784e-05,
+      "loss": 4.7185,
+      "step": 8082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.152365083773026e-05,
+      "loss": 0.5392,
+      "step": 8084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1519022493751735e-05,
+      "loss": 7.2309,
+      "step": 8086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1514394149773214e-05,
+      "loss": 0.977,
+      "step": 8088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1509765805794686e-05,
+      "loss": 1.7152,
+      "step": 8090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1505137461816165e-05,
+      "loss": 5.0554,
+      "step": 8092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.150050911783764e-05,
+      "loss": 1.8791,
+      "step": 8094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1495880773859116e-05,
+      "loss": 0.8723,
+      "step": 8096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.149125242988059e-05,
+      "loss": 1.4803,
+      "step": 8098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1486624085902067e-05,
+      "loss": 1.5806,
+      "step": 8100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1481995741923545e-05,
+      "loss": 0.0017,
+      "step": 8102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.147736739794502e-05,
+      "loss": 3.2618,
+      "step": 8104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1472739053966496e-05,
+      "loss": 2.5934,
+      "step": 8106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.146811070998797e-05,
+      "loss": 0.0591,
+      "step": 8108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.146348236600945e-05,
+      "loss": 3.1082,
+      "step": 8110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.145885402203092e-05,
+      "loss": 0.0131,
+      "step": 8112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.14542256780524e-05,
+      "loss": 0.4143,
+      "step": 8114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.144959733407387e-05,
+      "loss": 2.0981,
+      "step": 8116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.144496899009535e-05,
+      "loss": 0.4566,
+      "step": 8118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.144034064611682e-05,
+      "loss": 3.3129,
+      "step": 8120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.14357123021383e-05,
+      "loss": 3.7437,
+      "step": 8122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.143108395815977e-05,
+      "loss": 2.561,
+      "step": 8124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.142645561418125e-05,
+      "loss": 0.3355,
+      "step": 8126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.142182727020272e-05,
+      "loss": 0.2487,
+      "step": 8128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.14171989262242e-05,
+      "loss": 5.1461,
+      "step": 8130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1412570582245673e-05,
+      "loss": 5.9181,
+      "step": 8132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.140794223826715e-05,
+      "loss": 0.0027,
+      "step": 8134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1403313894288624e-05,
+      "loss": 0.9831,
+      "step": 8136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.13986855503101e-05,
+      "loss": 3.2638,
+      "step": 8138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.139405720633158e-05,
+      "loss": 1.8449,
+      "step": 8140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1389428862353054e-05,
+      "loss": 5.4895,
+      "step": 8142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.138480051837453e-05,
+      "loss": 3.4913,
+      "step": 8144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1380172174396005e-05,
+      "loss": 4.3024,
+      "step": 8146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.137554383041748e-05,
+      "loss": 2.2776,
+      "step": 8148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.137091548643895e-05,
+      "loss": 0.5822,
+      "step": 8150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.136628714246043e-05,
+      "loss": 0.7959,
+      "step": 8152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.13616587984819e-05,
+      "loss": 0.5333,
+      "step": 8154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.135703045450338e-05,
+      "loss": 2.6373,
+      "step": 8156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.135240211052485e-05,
+      "loss": 1.6578,
+      "step": 8158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.134777376654633e-05,
+      "loss": 0.0388,
+      "step": 8160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.13431454225678e-05,
+      "loss": 0.0395,
+      "step": 8162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.133851707858928e-05,
+      "loss": 0.0823,
+      "step": 8164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.133388873461076e-05,
+      "loss": 1.3693,
+      "step": 8166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.132926039063223e-05,
+      "loss": 0.0003,
+      "step": 8168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.132463204665371e-05,
+      "loss": 2.4051,
+      "step": 8170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.132000370267518e-05,
+      "loss": 1.7586,
+      "step": 8172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.131537535869666e-05,
+      "loss": 8.9745,
+      "step": 8174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.131074701471813e-05,
+      "loss": 2.4175,
+      "step": 8176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.130611867073961e-05,
+      "loss": 3.6859,
+      "step": 8178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1301490326761084e-05,
+      "loss": 0.1477,
+      "step": 8180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.129686198278256e-05,
+      "loss": 1.0666,
+      "step": 8182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1292233638804034e-05,
+      "loss": 0.5813,
+      "step": 8184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.128760529482551e-05,
+      "loss": 0.2749,
+      "step": 8186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1282976950846985e-05,
+      "loss": 5.8948,
+      "step": 8188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1278348606868464e-05,
+      "loss": 5.304,
+      "step": 8190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1273720262889936e-05,
+      "loss": 3.0641,
+      "step": 8192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1269091918911415e-05,
+      "loss": 0.0006,
+      "step": 8194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.126446357493289e-05,
+      "loss": 2.217,
+      "step": 8196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1259835230954366e-05,
+      "loss": 2.808,
+      "step": 8198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.125520688697584e-05,
+      "loss": 0.1423,
+      "step": 8200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.125057854299732e-05,
+      "loss": 5.0532,
+      "step": 8202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.124595019901879e-05,
+      "loss": 0.0093,
+      "step": 8204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.124132185504027e-05,
+      "loss": 1.9001,
+      "step": 8206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1236693511061746e-05,
+      "loss": 1.2191,
+      "step": 8208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.123206516708322e-05,
+      "loss": 7.868,
+      "step": 8210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.12274368231047e-05,
+      "loss": 4.8843,
+      "step": 8212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.122280847912617e-05,
+      "loss": 0.1155,
+      "step": 8214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.121818013514765e-05,
+      "loss": 0.2419,
+      "step": 8216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.121355179116912e-05,
+      "loss": 0.7358,
+      "step": 8218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.12089234471906e-05,
+      "loss": 3.0502,
+      "step": 8220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.120429510321207e-05,
+      "loss": 0.1567,
+      "step": 8222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.119966675923355e-05,
+      "loss": 5.584,
+      "step": 8224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.119503841525502e-05,
+      "loss": 0.4278,
+      "step": 8226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.11904100712765e-05,
+      "loss": 0.0003,
+      "step": 8228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.118578172729797e-05,
+      "loss": 5.1811,
+      "step": 8230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.118115338331945e-05,
+      "loss": 2.6023,
+      "step": 8232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1176525039340923e-05,
+      "loss": 0.0322,
+      "step": 8234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.11718966953624e-05,
+      "loss": 1.6061,
+      "step": 8236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1167268351383874e-05,
+      "loss": 2.189,
+      "step": 8238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.116264000740535e-05,
+      "loss": 0.0011,
+      "step": 8240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1158011663426825e-05,
+      "loss": 1.8465,
+      "step": 8242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1153383319448304e-05,
+      "loss": 0.4413,
+      "step": 8244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.114875497546978e-05,
+      "loss": 4.1425,
+      "step": 8246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1144126631491255e-05,
+      "loss": 0.0013,
+      "step": 8248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1139498287512734e-05,
+      "loss": 0.1166,
+      "step": 8250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1134869943534206e-05,
+      "loss": 3.2647,
+      "step": 8252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1130241599555684e-05,
+      "loss": 0.5009,
+      "step": 8254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1125613255577156e-05,
+      "loss": 0.0479,
+      "step": 8256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1120984911598635e-05,
+      "loss": 1.7526,
+      "step": 8258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.111635656762011e-05,
+      "loss": 5.7401,
+      "step": 8260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1111728223641586e-05,
+      "loss": 2.0462,
+      "step": 8262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.110709987966306e-05,
+      "loss": 2.2026,
+      "step": 8264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.110247153568454e-05,
+      "loss": 0.2488,
+      "step": 8266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.109784319170601e-05,
+      "loss": 3.4769,
+      "step": 8268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.109321484772749e-05,
+      "loss": 5.2134,
+      "step": 8270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.108858650374896e-05,
+      "loss": 1.5184,
+      "step": 8272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.108395815977044e-05,
+      "loss": 0.9441,
+      "step": 8274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.107932981579191e-05,
+      "loss": 0.2435,
+      "step": 8276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.107470147181339e-05,
+      "loss": 2.262,
+      "step": 8278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.107007312783486e-05,
+      "loss": 1.4654,
+      "step": 8280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.106544478385634e-05,
+      "loss": 2.0017,
+      "step": 8282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.106081643987781e-05,
+      "loss": 4.1003,
+      "step": 8284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.105618809589929e-05,
+      "loss": 6.4489,
+      "step": 8286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.105155975192077e-05,
+      "loss": 0.7611,
+      "step": 8288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.104693140794224e-05,
+      "loss": 1.8103,
+      "step": 8290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.104230306396372e-05,
+      "loss": 0.6592,
+      "step": 8292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1037674719985186e-05,
+      "loss": 6.3593,
+      "step": 8294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1033046376006665e-05,
+      "loss": 1.6152,
+      "step": 8296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.102841803202814e-05,
+      "loss": 3.1923,
+      "step": 8298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1023789688049616e-05,
+      "loss": 5.0264,
+      "step": 8300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.101916134407109e-05,
+      "loss": 1.48,
+      "step": 8302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.101453300009257e-05,
+      "loss": 3.8563,
+      "step": 8304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.100990465611404e-05,
+      "loss": 2.5706,
+      "step": 8306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.100527631213552e-05,
+      "loss": 0.5912,
+      "step": 8308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.100064796815699e-05,
+      "loss": 1.1973,
+      "step": 8310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.099601962417847e-05,
+      "loss": 1.9169,
+      "step": 8312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.099139128019995e-05,
+      "loss": 1.4483,
+      "step": 8314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.098676293622142e-05,
+      "loss": 5.5765,
+      "step": 8316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.09821345922429e-05,
+      "loss": 3.1531,
+      "step": 8318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.097750624826437e-05,
+      "loss": 0.3868,
+      "step": 8320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.097287790428585e-05,
+      "loss": 3.3188,
+      "step": 8322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.096824956030732e-05,
+      "loss": 0.0283,
+      "step": 8324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.09636212163288e-05,
+      "loss": 0.3323,
+      "step": 8326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.095899287235027e-05,
+      "loss": 4.2733,
+      "step": 8328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.095436452837175e-05,
+      "loss": 4.0463,
+      "step": 8330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.094973618439322e-05,
+      "loss": 0.3585,
+      "step": 8332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.09451078404147e-05,
+      "loss": 5.7527,
+      "step": 8334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0940479496436173e-05,
+      "loss": 2.2224,
+      "step": 8336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.093585115245765e-05,
+      "loss": 3.3552,
+      "step": 8338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0931222808479124e-05,
+      "loss": 2.3438,
+      "step": 8340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.09265944645006e-05,
+      "loss": 1.8997,
+      "step": 8342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0921966120522075e-05,
+      "loss": 0.9363,
+      "step": 8344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0917337776543554e-05,
+      "loss": 2.2877,
+      "step": 8346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0912709432565026e-05,
+      "loss": 1.2183,
+      "step": 8348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0908081088586505e-05,
+      "loss": 0.8037,
+      "step": 8350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.090345274460798e-05,
+      "loss": 1.7496,
+      "step": 8352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0898824400629456e-05,
+      "loss": 6.7742,
+      "step": 8354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0894196056650935e-05,
+      "loss": 2.9634,
+      "step": 8356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0889567712672407e-05,
+      "loss": 4.8109,
+      "step": 8358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0884939368693885e-05,
+      "loss": 1.7785,
+      "step": 8360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.088031102471536e-05,
+      "loss": 4.4072,
+      "step": 8362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0875682680736836e-05,
+      "loss": 2.5084,
+      "step": 8364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.087105433675831e-05,
+      "loss": 1.679,
+      "step": 8366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.086642599277979e-05,
+      "loss": 2.9228,
+      "step": 8368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.086179764880126e-05,
+      "loss": 4.0702,
+      "step": 8370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.085716930482274e-05,
+      "loss": 0.9584,
+      "step": 8372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.085254096084421e-05,
+      "loss": 2.5312,
+      "step": 8374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.084791261686569e-05,
+      "loss": 2.9919,
+      "step": 8376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.084328427288716e-05,
+      "loss": 2.0623,
+      "step": 8378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.083865592890864e-05,
+      "loss": 0.3563,
+      "step": 8380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.083402758493011e-05,
+      "loss": 0.9352,
+      "step": 8382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.082939924095159e-05,
+      "loss": 0.0363,
+      "step": 8384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.082477089697306e-05,
+      "loss": 4.4284,
+      "step": 8386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.082014255299454e-05,
+      "loss": 6.0555,
+      "step": 8388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.081551420901601e-05,
+      "loss": 1.7491,
+      "step": 8390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.081088586503749e-05,
+      "loss": 4.2498,
+      "step": 8392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.080625752105897e-05,
+      "loss": 6.2035,
+      "step": 8394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.080162917708044e-05,
+      "loss": 0.2129,
+      "step": 8396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.079700083310192e-05,
+      "loss": 1.7508,
+      "step": 8398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0792372489123394e-05,
+      "loss": 1.4388,
+      "step": 8400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.078774414514487e-05,
+      "loss": 1.2908,
+      "step": 8402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0783115801166345e-05,
+      "loss": 0.7245,
+      "step": 8404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0778487457187824e-05,
+      "loss": 0.912,
+      "step": 8406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0773859113209296e-05,
+      "loss": 5.0618,
+      "step": 8408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.0153,
+      "step": 8410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0764602425252246e-05,
+      "loss": 0.2026,
+      "step": 8412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0759974081273725e-05,
+      "loss": 2.4413,
+      "step": 8414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.07553457372952e-05,
+      "loss": 2.3155,
+      "step": 8416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0750717393316676e-05,
+      "loss": 2.0893,
+      "step": 8418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.074608904933815e-05,
+      "loss": 5.1146,
+      "step": 8420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.074146070535963e-05,
+      "loss": 6.236,
+      "step": 8422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.07368323613811e-05,
+      "loss": 3.1347,
+      "step": 8424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.073220401740258e-05,
+      "loss": 2.2774,
+      "step": 8426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.072757567342405e-05,
+      "loss": 4.4451,
+      "step": 8428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.072294732944553e-05,
+      "loss": 6.6853,
+      "step": 8430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0718318985467e-05,
+      "loss": 1.6249,
+      "step": 8432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.071369064148848e-05,
+      "loss": 1.3372,
+      "step": 8434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.070906229750996e-05,
+      "loss": 0.5749,
+      "step": 8436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0704433953531424e-05,
+      "loss": 0.4462,
+      "step": 8438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.06998056095529e-05,
+      "loss": 0.9555,
+      "step": 8440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0695177265574374e-05,
+      "loss": 0.7869,
+      "step": 8442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.069054892159585e-05,
+      "loss": 3.6525,
+      "step": 8444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0685920577617325e-05,
+      "loss": 3.6249,
+      "step": 8446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0681292233638804e-05,
+      "loss": 0.0006,
+      "step": 8448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0676663889660276e-05,
+      "loss": 1.2922,
+      "step": 8450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0672035545681755e-05,
+      "loss": 1.0289,
+      "step": 8452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.066740720170323e-05,
+      "loss": 6.1991,
+      "step": 8454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0662778857724706e-05,
+      "loss": 0.3071,
+      "step": 8456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.065815051374618e-05,
+      "loss": 2.1744,
+      "step": 8458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0653522169767657e-05,
+      "loss": 1.9806,
+      "step": 8460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0648893825789135e-05,
+      "loss": 0.5833,
+      "step": 8462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.064426548181061e-05,
+      "loss": 3.3545,
+      "step": 8464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0639637137832086e-05,
+      "loss": 1.5395,
+      "step": 8466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.063500879385356e-05,
+      "loss": 2.1006,
+      "step": 8468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.063038044987504e-05,
+      "loss": 0.1467,
+      "step": 8470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.062575210589651e-05,
+      "loss": 2.8177,
+      "step": 8472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.062112376191799e-05,
+      "loss": 1.004,
+      "step": 8474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.061649541793946e-05,
+      "loss": 1.4883,
+      "step": 8476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.061186707396094e-05,
+      "loss": 0.5784,
+      "step": 8478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.060723872998241e-05,
+      "loss": 0.006,
+      "step": 8480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.060261038600389e-05,
+      "loss": 0.0005,
+      "step": 8482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.059798204202536e-05,
+      "loss": 0.0816,
+      "step": 8484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.059335369804684e-05,
+      "loss": 0.0325,
+      "step": 8486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.058872535406831e-05,
+      "loss": 1.665,
+      "step": 8488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.058409701008979e-05,
+      "loss": 6.3894,
+      "step": 8490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.057946866611126e-05,
+      "loss": 0.0003,
+      "step": 8492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.057484032213274e-05,
+      "loss": 2.6149,
+      "step": 8494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0570211978154214e-05,
+      "loss": 0.9966,
+      "step": 8496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.056558363417569e-05,
+      "loss": 4.2451,
+      "step": 8498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.056095529019717e-05,
+      "loss": 2.0741,
+      "step": 8500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0556326946218644e-05,
+      "loss": 0.0004,
+      "step": 8502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.055169860224012e-05,
+      "loss": 1.3691,
+      "step": 8504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0547070258261595e-05,
+      "loss": 2.3849,
+      "step": 8506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0542441914283074e-05,
+      "loss": 1.4019,
+      "step": 8508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0537813570304546e-05,
+      "loss": 0.0646,
+      "step": 8510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0533185226326024e-05,
+      "loss": 2.5545,
+      "step": 8512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0528556882347496e-05,
+      "loss": 1.0982,
+      "step": 8514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0523928538368975e-05,
+      "loss": 7.2217,
+      "step": 8516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.051930019439045e-05,
+      "loss": 1.1519,
+      "step": 8518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0514671850411926e-05,
+      "loss": 0.9134,
+      "step": 8520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.05100435064334e-05,
+      "loss": 2.7636,
+      "step": 8522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0505415162454877e-05,
+      "loss": 0.0745,
+      "step": 8524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0500786818476352e-05,
+      "loss": 3.6174,
+      "step": 8526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0496158474497828e-05,
+      "loss": 4.757,
+      "step": 8528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0491530130519303e-05,
+      "loss": 2.1492,
+      "step": 8530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.048690178654078e-05,
+      "loss": 0.0002,
+      "step": 8532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0482273442562254e-05,
+      "loss": 3.0494,
+      "step": 8534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.047764509858373e-05,
+      "loss": 3.8122,
+      "step": 8536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0473016754605205e-05,
+      "loss": 1.2335,
+      "step": 8538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.046838841062668e-05,
+      "loss": 2.2388,
+      "step": 8540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0463760066648156e-05,
+      "loss": 0.8869,
+      "step": 8542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.045913172266963e-05,
+      "loss": 3.7308,
+      "step": 8544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0454503378691107e-05,
+      "loss": 0.2315,
+      "step": 8546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0449875034712582e-05,
+      "loss": 3.8689,
+      "step": 8548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0445246690734057e-05,
+      "loss": 2.5855,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0440618346755533e-05,
+      "loss": 3.0806,
+      "step": 8552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.043599000277701e-05,
+      "loss": 1.4234,
+      "step": 8554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0431361658798484e-05,
+      "loss": 3.9798,
+      "step": 8556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.042673331481996e-05,
+      "loss": 4.0061,
+      "step": 8558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0422104970841438e-05,
+      "loss": 4.0273,
+      "step": 8560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0417476626862913e-05,
+      "loss": 1.7658,
+      "step": 8562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.041284828288439e-05,
+      "loss": 1.1159,
+      "step": 8564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0408219938905864e-05,
+      "loss": 1.5655,
+      "step": 8566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.040359159492734e-05,
+      "loss": 4.1492,
+      "step": 8568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0398963250948815e-05,
+      "loss": 1.8282,
+      "step": 8570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.039433490697029e-05,
+      "loss": 3.0555,
+      "step": 8572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0389706562991766e-05,
+      "loss": 2.3157,
+      "step": 8574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.038507821901324e-05,
+      "loss": 2.9472,
+      "step": 8576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0380449875034717e-05,
+      "loss": 1.0291,
+      "step": 8578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0375821531056192e-05,
+      "loss": 0.8681,
+      "step": 8580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0371193187077668e-05,
+      "loss": 2.9629,
+      "step": 8582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0366564843099136e-05,
+      "loss": 3.7028,
+      "step": 8584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0361936499120615e-05,
+      "loss": 1.1608,
+      "step": 8586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.035730815514209e-05,
+      "loss": 1.9605,
+      "step": 8588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0352679811163566e-05,
+      "loss": 4.3788,
+      "step": 8590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.034805146718504e-05,
+      "loss": 2.5573,
+      "step": 8592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0343423123206517e-05,
+      "loss": 5.0749,
+      "step": 8594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0338794779227992e-05,
+      "loss": 2.3817,
+      "step": 8596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0334166435249468e-05,
+      "loss": 1.3812,
+      "step": 8598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0329538091270943e-05,
+      "loss": 1.0008,
+      "step": 8600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.032490974729242e-05,
+      "loss": 0.776,
+      "step": 8602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0320281403313894e-05,
+      "loss": 0.9474,
+      "step": 8604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.031565305933537e-05,
+      "loss": 1.5358,
+      "step": 8606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0311024715356845e-05,
+      "loss": 0.4011,
+      "step": 8608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.030639637137832e-05,
+      "loss": 0.6522,
+      "step": 8610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0301768027399796e-05,
+      "loss": 0.2305,
+      "step": 8612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.029713968342127e-05,
+      "loss": 0.5493,
+      "step": 8614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0292511339442746e-05,
+      "loss": 6.2125,
+      "step": 8616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0287882995464222e-05,
+      "loss": 1.9507,
+      "step": 8618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0283254651485697e-05,
+      "loss": 0.006,
+      "step": 8620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0278626307507173e-05,
+      "loss": 0.5366,
+      "step": 8622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0273997963528648e-05,
+      "loss": 2.3751,
+      "step": 8624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0269369619550127e-05,
+      "loss": 0.0048,
+      "step": 8626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0264741275571602e-05,
+      "loss": 6.6239,
+      "step": 8628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0260112931593078e-05,
+      "loss": 0.0148,
+      "step": 8630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0255484587614553e-05,
+      "loss": 3.9663,
+      "step": 8632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.025085624363603e-05,
+      "loss": 8.3482,
+      "step": 8634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0246227899657504e-05,
+      "loss": 4.7238,
+      "step": 8636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.024159955567898e-05,
+      "loss": 7.1077,
+      "step": 8638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0236971211700455e-05,
+      "loss": 2.7613,
+      "step": 8640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.023234286772193e-05,
+      "loss": 5.9056,
+      "step": 8642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0227714523743406e-05,
+      "loss": 0.002,
+      "step": 8644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.022308617976488e-05,
+      "loss": 0.5234,
+      "step": 8646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0218457835786357e-05,
+      "loss": 2.3803,
+      "step": 8648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0213829491807832e-05,
+      "loss": 1.6532,
+      "step": 8650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0209201147829308e-05,
+      "loss": 2.0159,
+      "step": 8652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0204572803850783e-05,
+      "loss": 1.7929,
+      "step": 8654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.019994445987226e-05,
+      "loss": 0.2551,
+      "step": 8656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0195316115893734e-05,
+      "loss": 1.0449,
+      "step": 8658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.019068777191521e-05,
+      "loss": 1.9425,
+      "step": 8660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0186059427936685e-05,
+      "loss": 1.2776,
+      "step": 8662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.018143108395816e-05,
+      "loss": 0.2275,
+      "step": 8664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0176802739979635e-05,
+      "loss": 0.6151,
+      "step": 8666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0172174396001114e-05,
+      "loss": 5.2266,
+      "step": 8668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.016754605202259e-05,
+      "loss": 2.0167,
+      "step": 8670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0162917708044065e-05,
+      "loss": 4.5322,
+      "step": 8672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.015828936406554e-05,
+      "loss": 2.6235,
+      "step": 8674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0153661020087016e-05,
+      "loss": 0.0141,
+      "step": 8676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.014903267610849e-05,
+      "loss": 4.3783,
+      "step": 8678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0144404332129967e-05,
+      "loss": 7.0674,
+      "step": 8680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0139775988151442e-05,
+      "loss": 2.242,
+      "step": 8682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0135147644172918e-05,
+      "loss": 2.026,
+      "step": 8684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0130519300194393e-05,
+      "loss": 0.5642,
+      "step": 8686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.012589095621587e-05,
+      "loss": 3.2346,
+      "step": 8688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0121262612237344e-05,
+      "loss": 1.7129,
+      "step": 8690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.011663426825882e-05,
+      "loss": 0.848,
+      "step": 8692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0112005924280295e-05,
+      "loss": 0.137,
+      "step": 8694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.010737758030177e-05,
+      "loss": 0.8701,
+      "step": 8696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0102749236323246e-05,
+      "loss": 4.2081,
+      "step": 8698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.009812089234472e-05,
+      "loss": 1.0684,
+      "step": 8700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0093492548366197e-05,
+      "loss": 0.9675,
+      "step": 8702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0088864204387672e-05,
+      "loss": 1.0502,
+      "step": 8704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0084235860409147e-05,
+      "loss": 3.2765,
+      "step": 8706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0079607516430626e-05,
+      "loss": 2.1821,
+      "step": 8708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.00749791724521e-05,
+      "loss": 1.1262,
+      "step": 8710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0070350828473577e-05,
+      "loss": 0.045,
+      "step": 8712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0065722484495052e-05,
+      "loss": 0.0179,
+      "step": 8714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0061094140516528e-05,
+      "loss": 0.9384,
+      "step": 8716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0056465796538003e-05,
+      "loss": 1.9537,
+      "step": 8718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.005183745255948e-05,
+      "loss": 3.5294,
+      "step": 8720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0047209108580954e-05,
+      "loss": 2.6424,
+      "step": 8722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.004258076460243e-05,
+      "loss": 1.122,
+      "step": 8724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0037952420623905e-05,
+      "loss": 4.2658,
+      "step": 8726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0033324076645374e-05,
+      "loss": 3.839,
+      "step": 8728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.002869573266685e-05,
+      "loss": 0.0003,
+      "step": 8730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0024067388688324e-05,
+      "loss": 0.5661,
+      "step": 8732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0019439044709803e-05,
+      "loss": 5.1147,
+      "step": 8734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.001481070073128e-05,
+      "loss": 4.5178,
+      "step": 8736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0010182356752754e-05,
+      "loss": 1.9922,
+      "step": 8738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.000555401277423e-05,
+      "loss": 2.6919,
+      "step": 8740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.0000925668795705e-05,
+      "loss": 6.1803,
+      "step": 8742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.999629732481718e-05,
+      "loss": 4.4818,
+      "step": 8744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9991668980838656e-05,
+      "loss": 0.92,
+      "step": 8746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998704063686013e-05,
+      "loss": 2.3221,
+      "step": 8748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982412292881607e-05,
+      "loss": 4.035,
+      "step": 8750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9977783948903082e-05,
+      "loss": 2.1704,
+      "step": 8752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9973155604924558e-05,
+      "loss": 0.3782,
+      "step": 8754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9968527260946033e-05,
+      "loss": 3.8531,
+      "step": 8756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.996389891696751e-05,
+      "loss": 3.9853,
+      "step": 8758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9959270572988984e-05,
+      "loss": 1.3531,
+      "step": 8760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.995464222901046e-05,
+      "loss": 2.8588,
+      "step": 8762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9950013885031935e-05,
+      "loss": 0.8375,
+      "step": 8764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.994538554105341e-05,
+      "loss": 0.0221,
+      "step": 8766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9940757197074886e-05,
+      "loss": 1.1433,
+      "step": 8768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.993612885309636e-05,
+      "loss": 1.965,
+      "step": 8770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9931500509117836e-05,
+      "loss": 2.3925,
+      "step": 8772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9926872165139315e-05,
+      "loss": 2.5807,
+      "step": 8774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.992224382116079e-05,
+      "loss": 2.6032,
+      "step": 8776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9917615477182266e-05,
+      "loss": 5.8284,
+      "step": 8778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.991298713320374e-05,
+      "loss": 2.644,
+      "step": 8780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9908358789225217e-05,
+      "loss": 0.5209,
+      "step": 8782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9903730445246692e-05,
+      "loss": 4.0014,
+      "step": 8784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9899102101268168e-05,
+      "loss": 2.3262,
+      "step": 8786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9894473757289643e-05,
+      "loss": 2.6605,
+      "step": 8788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.988984541331112e-05,
+      "loss": 0.9207,
+      "step": 8790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9885217069332594e-05,
+      "loss": 1.551,
+      "step": 8792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.988058872535407e-05,
+      "loss": 0.1716,
+      "step": 8794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9875960381375545e-05,
+      "loss": 0.086,
+      "step": 8796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.987133203739702e-05,
+      "loss": 2.2918,
+      "step": 8798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9866703693418496e-05,
+      "loss": 0.9051,
+      "step": 8800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.986207534943997e-05,
+      "loss": 1.5911,
+      "step": 8802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9857447005461447e-05,
+      "loss": 0.2114,
+      "step": 8804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9852818661482922e-05,
+      "loss": 1.3804,
+      "step": 8806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9848190317504397e-05,
+      "loss": 0.0005,
+      "step": 8808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9843561973525873e-05,
+      "loss": 0.5363,
+      "step": 8810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9838933629547348e-05,
+      "loss": 1.3634,
+      "step": 8812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9834305285568827e-05,
+      "loss": 4.1716,
+      "step": 8814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9829676941590302e-05,
+      "loss": 0.9799,
+      "step": 8816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9825048597611778e-05,
+      "loss": 0.2878,
+      "step": 8818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9820420253633253e-05,
+      "loss": 0.9264,
+      "step": 8820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.981579190965473e-05,
+      "loss": 0.0007,
+      "step": 8822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9811163565676204e-05,
+      "loss": 5.956,
+      "step": 8824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.980653522169768e-05,
+      "loss": 5.9671,
+      "step": 8826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9801906877719155e-05,
+      "loss": 0.0964,
+      "step": 8828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.979727853374063e-05,
+      "loss": 0.0004,
+      "step": 8830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9792650189762106e-05,
+      "loss": 0.6169,
+      "step": 8832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.978802184578358e-05,
+      "loss": 0.0013,
+      "step": 8834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9783393501805057e-05,
+      "loss": 1.0167,
+      "step": 8836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9778765157826532e-05,
+      "loss": 2.9469,
+      "step": 8838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9774136813848008e-05,
+      "loss": 0.0006,
+      "step": 8840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9769508469869483e-05,
+      "loss": 1.0241,
+      "step": 8842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.976488012589096e-05,
+      "loss": 8.7453,
+      "step": 8844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9760251781912434e-05,
+      "loss": 1.7243,
+      "step": 8846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.975562343793391e-05,
+      "loss": 2.9527,
+      "step": 8848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9750995093955385e-05,
+      "loss": 1.2472,
+      "step": 8850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.974636674997686e-05,
+      "loss": 0.0023,
+      "step": 8852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.974173840599834e-05,
+      "loss": 3.3608,
+      "step": 8854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9737110062019814e-05,
+      "loss": 0.2277,
+      "step": 8856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.973248171804129e-05,
+      "loss": 5.9623,
+      "step": 8858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9727853374062765e-05,
+      "loss": 0.0256,
+      "step": 8860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.972322503008424e-05,
+      "loss": 1.715,
+      "step": 8862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9718596686105716e-05,
+      "loss": 0.1588,
+      "step": 8864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.971396834212719e-05,
+      "loss": 1.5474,
+      "step": 8866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9709339998148667e-05,
+      "loss": 4.8032,
+      "step": 8868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9704711654170142e-05,
+      "loss": 4.3789,
+      "step": 8870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9700083310191618e-05,
+      "loss": 0.0108,
+      "step": 8872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9695454966213086e-05,
+      "loss": 1.2991,
+      "step": 8874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9690826622234562e-05,
+      "loss": 0.1839,
+      "step": 8876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9686198278256037e-05,
+      "loss": 0.3276,
+      "step": 8878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9681569934277516e-05,
+      "loss": 2.9986,
+      "step": 8880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.967694159029899e-05,
+      "loss": 2.5504,
+      "step": 8882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9672313246320467e-05,
+      "loss": 1.7614,
+      "step": 8884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9667684902341942e-05,
+      "loss": 0.9056,
+      "step": 8886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9663056558363418e-05,
+      "loss": 3.0661,
+      "step": 8888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9658428214384893e-05,
+      "loss": 0.0546,
+      "step": 8890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.965379987040637e-05,
+      "loss": 0.6598,
+      "step": 8892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9649171526427844e-05,
+      "loss": 1.5218,
+      "step": 8894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.964454318244932e-05,
+      "loss": 0.5795,
+      "step": 8896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9639914838470795e-05,
+      "loss": 5.3184,
+      "step": 8898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.963528649449227e-05,
+      "loss": 2.5297,
+      "step": 8900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9630658150513746e-05,
+      "loss": 1.7257,
+      "step": 8902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.962602980653522e-05,
+      "loss": 1.3044,
+      "step": 8904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9621401462556697e-05,
+      "loss": 1.6708,
+      "step": 8906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9616773118578172e-05,
+      "loss": 0.9601,
+      "step": 8908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9612144774599647e-05,
+      "loss": 1.4011,
+      "step": 8910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9607516430621123e-05,
+      "loss": 1.8843,
+      "step": 8912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9602888086642598e-05,
+      "loss": 2.9137,
+      "step": 8914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9598259742664074e-05,
+      "loss": 2.0101,
+      "step": 8916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.959363139868555e-05,
+      "loss": 2.5016,
+      "step": 8918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9589003054707028e-05,
+      "loss": 4.2511,
+      "step": 8920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9584374710728503e-05,
+      "loss": 4.6908,
+      "step": 8922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.957974636674998e-05,
+      "loss": 5.4428,
+      "step": 8924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9575118022771454e-05,
+      "loss": 0.1247,
+      "step": 8926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.957048967879293e-05,
+      "loss": 1.5471,
+      "step": 8928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9565861334814405e-05,
+      "loss": 1.0406,
+      "step": 8930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.956123299083588e-05,
+      "loss": 0.2986,
+      "step": 8932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9556604646857356e-05,
+      "loss": 1.3935,
+      "step": 8934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.955197630287883e-05,
+      "loss": 3.3782,
+      "step": 8936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9547347958900307e-05,
+      "loss": 1.3272,
+      "step": 8938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9542719614921782e-05,
+      "loss": 2.7043,
+      "step": 8940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9538091270943258e-05,
+      "loss": 2.0393,
+      "step": 8942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9533462926964733e-05,
+      "loss": 0.0067,
+      "step": 8944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.952883458298621e-05,
+      "loss": 6.0855,
+      "step": 8946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9524206239007684e-05,
+      "loss": 3.7209,
+      "step": 8948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951957789502916e-05,
+      "loss": 2.4002,
+      "step": 8950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9514949551050635e-05,
+      "loss": 2.6307,
+      "step": 8952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.951032120707211e-05,
+      "loss": 3.7952,
+      "step": 8954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9505692863093586e-05,
+      "loss": 0.002,
+      "step": 8956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.950106451911506e-05,
+      "loss": 3.7827,
+      "step": 8958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9496436175136536e-05,
+      "loss": 0.3493,
+      "step": 8960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9491807831158015e-05,
+      "loss": 0.4392,
+      "step": 8962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.948717948717949e-05,
+      "loss": 1.1464,
+      "step": 8964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9482551143200966e-05,
+      "loss": 0.0856,
+      "step": 8966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.947792279922244e-05,
+      "loss": 6.4099,
+      "step": 8968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9473294455243917e-05,
+      "loss": 2.5856,
+      "step": 8970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9468666111265392e-05,
+      "loss": 5.3615,
+      "step": 8972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9464037767286868e-05,
+      "loss": 0.7539,
+      "step": 8974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9459409423308343e-05,
+      "loss": 4.4653,
+      "step": 8976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.945478107932982e-05,
+      "loss": 1.0239,
+      "step": 8978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9450152735351294e-05,
+      "loss": 0.0162,
+      "step": 8980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.944552439137277e-05,
+      "loss": 1.1272,
+      "step": 8982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9440896047394245e-05,
+      "loss": 1.8814,
+      "step": 8984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.943626770341572e-05,
+      "loss": 3.061,
+      "step": 8986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9431639359437196e-05,
+      "loss": 1.3125,
+      "step": 8988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.942701101545867e-05,
+      "loss": 0.6489,
+      "step": 8990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9422382671480147e-05,
+      "loss": 4.5921,
+      "step": 8992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9417754327501622e-05,
+      "loss": 1.4663,
+      "step": 8994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9413125983523097e-05,
+      "loss": 2.4804,
+      "step": 8996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9408497639544573e-05,
+      "loss": 1.4259,
+      "step": 8998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.940386929556605e-05,
+      "loss": 5.108,
+      "step": 9000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9399240951587527e-05,
+      "loss": 2.2584,
+      "step": 9002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9394612607609003e-05,
+      "loss": 2.4195,
+      "step": 9004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9389984263630478e-05,
+      "loss": 1.5064,
+      "step": 9006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9385355919651953e-05,
+      "loss": 2.1599,
+      "step": 9008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.938072757567343e-05,
+      "loss": 4.3946,
+      "step": 9010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9376099231694904e-05,
+      "loss": 1.8067,
+      "step": 9012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.937147088771638e-05,
+      "loss": 1.6209,
+      "step": 9014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9366842543737855e-05,
+      "loss": 2.39,
+      "step": 9016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9362214199759324e-05,
+      "loss": 1.8602,
+      "step": 9018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.93575858557808e-05,
+      "loss": 0.3118,
+      "step": 9020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9352957511802275e-05,
+      "loss": 2.661,
+      "step": 9022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.934832916782375e-05,
+      "loss": 1.5692,
+      "step": 9024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9343700823845225e-05,
+      "loss": 3.7413,
+      "step": 9026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9339072479866704e-05,
+      "loss": 0.7714,
+      "step": 9028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.933444413588818e-05,
+      "loss": 0.0004,
+      "step": 9030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9329815791909655e-05,
+      "loss": 2.6258,
+      "step": 9032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.932518744793113e-05,
+      "loss": 0.2485,
+      "step": 9034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9320559103952606e-05,
+      "loss": 2.1368,
+      "step": 9036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.931593075997408e-05,
+      "loss": 2.1491,
+      "step": 9038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9311302415995557e-05,
+      "loss": 1.2984,
+      "step": 9040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9306674072017032e-05,
+      "loss": 0.317,
+      "step": 9042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9302045728038508e-05,
+      "loss": 0.468,
+      "step": 9044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9297417384059983e-05,
+      "loss": 0.1246,
+      "step": 9046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.929278904008146e-05,
+      "loss": 4.8425,
+      "step": 9048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9288160696102934e-05,
+      "loss": 4.982,
+      "step": 9050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.928353235212441e-05,
+      "loss": 2.0127,
+      "step": 9052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9278904008145885e-05,
+      "loss": 3.79,
+      "step": 9054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.927427566416736e-05,
+      "loss": 2.6994,
+      "step": 9056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9269647320188836e-05,
+      "loss": 0.4193,
+      "step": 9058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.926501897621031e-05,
+      "loss": 5.9747,
+      "step": 9060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9260390632231786e-05,
+      "loss": 1.2185,
+      "step": 9062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9255762288253262e-05,
+      "loss": 2.3477,
+      "step": 9064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9251133944274737e-05,
+      "loss": 1.6976,
+      "step": 9066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9246505600296216e-05,
+      "loss": 1.2334,
+      "step": 9068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.924187725631769e-05,
+      "loss": 1.019,
+      "step": 9070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9237248912339167e-05,
+      "loss": 1.6281,
+      "step": 9072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9232620568360642e-05,
+      "loss": 3.9281,
+      "step": 9074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9227992224382118e-05,
+      "loss": 5.6147,
+      "step": 9076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9223363880403593e-05,
+      "loss": 1.2166,
+      "step": 9078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.921873553642507e-05,
+      "loss": 1.6301,
+      "step": 9080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9214107192446544e-05,
+      "loss": 2.411,
+      "step": 9082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.920947884846802e-05,
+      "loss": 4.9103,
+      "step": 9084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9204850504489495e-05,
+      "loss": 2.6835,
+      "step": 9086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.920022216051097e-05,
+      "loss": 2.8274,
+      "step": 9088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9195593816532446e-05,
+      "loss": 1.2848,
+      "step": 9090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.919096547255392e-05,
+      "loss": 2.9705,
+      "step": 9092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9186337128575397e-05,
+      "loss": 1.4791,
+      "step": 9094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9181708784596872e-05,
+      "loss": 1.8438,
+      "step": 9096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9177080440618348e-05,
+      "loss": 2.6168,
+      "step": 9098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9172452096639823e-05,
+      "loss": 0.4934,
+      "step": 9100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.91678237526613e-05,
+      "loss": 1.5785,
+      "step": 9102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9163195408682774e-05,
+      "loss": 2.1452,
+      "step": 9104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.915856706470425e-05,
+      "loss": 2.3342,
+      "step": 9106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9153938720725728e-05,
+      "loss": 1.5154,
+      "step": 9108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9149310376747203e-05,
+      "loss": 2.5178,
+      "step": 9110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.914468203276868e-05,
+      "loss": 3.6622,
+      "step": 9112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9140053688790154e-05,
+      "loss": 0.9573,
+      "step": 9114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.913542534481163e-05,
+      "loss": 1.0877,
+      "step": 9116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9130797000833105e-05,
+      "loss": 3.2407,
+      "step": 9118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.912616865685458e-05,
+      "loss": 1.8821,
+      "step": 9120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9121540312876056e-05,
+      "loss": 1.68,
+      "step": 9122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.911691196889753e-05,
+      "loss": 5.2975,
+      "step": 9124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9112283624919007e-05,
+      "loss": 8.881,
+      "step": 9126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9107655280940482e-05,
+      "loss": 5.7818,
+      "step": 9128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9103026936961958e-05,
+      "loss": 0.1531,
+      "step": 9130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9098398592983433e-05,
+      "loss": 4.9098,
+      "step": 9132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.909377024900491e-05,
+      "loss": 1.1368,
+      "step": 9134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9089141905026384e-05,
+      "loss": 2.4192,
+      "step": 9136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.908451356104786e-05,
+      "loss": 1.3409,
+      "step": 9138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9079885217069335e-05,
+      "loss": 1.8571,
+      "step": 9140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.907525687309081e-05,
+      "loss": 1.8307,
+      "step": 9142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9070628529112286e-05,
+      "loss": 0.0022,
+      "step": 9144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.906600018513376e-05,
+      "loss": 2.6491,
+      "step": 9146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.906137184115524e-05,
+      "loss": 0.1569,
+      "step": 9148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9056743497176715e-05,
+      "loss": 0.0622,
+      "step": 9150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.905211515319819e-05,
+      "loss": 0.0025,
+      "step": 9152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9047486809219666e-05,
+      "loss": 1.4535,
+      "step": 9154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.904285846524114e-05,
+      "loss": 1.0994,
+      "step": 9156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9038230121262617e-05,
+      "loss": 3.8168,
+      "step": 9158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9033601777284092e-05,
+      "loss": 2.6239,
+      "step": 9160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9028973433305568e-05,
+      "loss": 4.714,
+      "step": 9162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9024345089327037e-05,
+      "loss": 3.7696,
+      "step": 9164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9019716745348512e-05,
+      "loss": 0.0001,
+      "step": 9166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9015088401369987e-05,
+      "loss": 0.3128,
+      "step": 9168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9010460057391463e-05,
+      "loss": 4.1992,
+      "step": 9170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9005831713412938e-05,
+      "loss": 6.061,
+      "step": 9172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9001203369434417e-05,
+      "loss": 1.3614,
+      "step": 9174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8996575025455892e-05,
+      "loss": 2.1659,
+      "step": 9176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8991946681477368e-05,
+      "loss": 0.8769,
+      "step": 9178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8987318337498843e-05,
+      "loss": 0.0429,
+      "step": 9180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.898268999352032e-05,
+      "loss": 2.3563,
+      "step": 9182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8978061649541794e-05,
+      "loss": 2.3059,
+      "step": 9184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.897343330556327e-05,
+      "loss": 0.525,
+      "step": 9186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8968804961584745e-05,
+      "loss": 1.9252,
+      "step": 9188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.896417661760622e-05,
+      "loss": 1.6974,
+      "step": 9190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8959548273627696e-05,
+      "loss": 0.9021,
+      "step": 9192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.895491992964917e-05,
+      "loss": 5.7722,
+      "step": 9194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8950291585670647e-05,
+      "loss": 0.4352,
+      "step": 9196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8945663241692122e-05,
+      "loss": 5.0691,
+      "step": 9198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8941034897713598e-05,
+      "loss": 0.5461,
+      "step": 9200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8936406553735073e-05,
+      "loss": 1.4365,
+      "step": 9202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.893177820975655e-05,
+      "loss": 0.0015,
+      "step": 9204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8927149865778024e-05,
+      "loss": 2.5322,
+      "step": 9206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.89225215217995e-05,
+      "loss": 0.0903,
+      "step": 9208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8917893177820975e-05,
+      "loss": 1.2638,
+      "step": 9210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.891326483384245e-05,
+      "loss": 2.2032,
+      "step": 9212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.890863648986393e-05,
+      "loss": 5.3867,
+      "step": 9214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8904008145885404e-05,
+      "loss": 1.7568,
+      "step": 9216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.889937980190688e-05,
+      "loss": 3.533,
+      "step": 9218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8894751457928355e-05,
+      "loss": 6.6463,
+      "step": 9220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.889012311394983e-05,
+      "loss": 3.0191,
+      "step": 9222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8885494769971306e-05,
+      "loss": 1.9215,
+      "step": 9224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.888086642599278e-05,
+      "loss": 1.2494,
+      "step": 9226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8876238082014257e-05,
+      "loss": 1.9135,
+      "step": 9228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8871609738035732e-05,
+      "loss": 1.7806,
+      "step": 9230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8866981394057208e-05,
+      "loss": 2.6866,
+      "step": 9232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8862353050078683e-05,
+      "loss": 2.7495,
+      "step": 9234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.885772470610016e-05,
+      "loss": 2.1765,
+      "step": 9236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8853096362121634e-05,
+      "loss": 1.2448,
+      "step": 9238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.884846801814311e-05,
+      "loss": 0.0092,
+      "step": 9240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8843839674164585e-05,
+      "loss": 0.4041,
+      "step": 9242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.883921133018606e-05,
+      "loss": 0.1657,
+      "step": 9244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8834582986207536e-05,
+      "loss": 0.425,
+      "step": 9246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.882995464222901e-05,
+      "loss": 0.0588,
+      "step": 9248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8825326298250487e-05,
+      "loss": 2.918,
+      "step": 9250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8820697954271962e-05,
+      "loss": 1.7307,
+      "step": 9252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.881606961029344e-05,
+      "loss": 0.1562,
+      "step": 9254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8811441266314916e-05,
+      "loss": 1.6641,
+      "step": 9256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.880681292233639e-05,
+      "loss": 2.4512,
+      "step": 9258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8802184578357867e-05,
+      "loss": 0.9938,
+      "step": 9260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8797556234379342e-05,
+      "loss": 0.0002,
+      "step": 9262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8792927890400818e-05,
+      "loss": 3.3522,
+      "step": 9264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8788299546422293e-05,
+      "loss": 0.0061,
+      "step": 9266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.878367120244377e-05,
+      "loss": 0.0002,
+      "step": 9268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8779042858465244e-05,
+      "loss": 0.0054,
+      "step": 9270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.877441451448672e-05,
+      "loss": 0.5236,
+      "step": 9272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8769786170508195e-05,
+      "loss": 1.187,
+      "step": 9274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.876515782652967e-05,
+      "loss": 2.3937,
+      "step": 9276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8760529482551146e-05,
+      "loss": 4.5175,
+      "step": 9278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.875590113857262e-05,
+      "loss": 4.9882,
+      "step": 9280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8751272794594097e-05,
+      "loss": 7.7477,
+      "step": 9282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8746644450615572e-05,
+      "loss": 5.8341,
+      "step": 9284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8742016106637048e-05,
+      "loss": 1.3651,
+      "step": 9286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8737387762658523e-05,
+      "loss": 0.8056,
+      "step": 9288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.873275941868e-05,
+      "loss": 1.4267,
+      "step": 9290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8728131074701474e-05,
+      "loss": 3.9288,
+      "step": 9292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.872350273072295e-05,
+      "loss": 0.8057,
+      "step": 9294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8718874386744428e-05,
+      "loss": 5.9239,
+      "step": 9296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8714246042765904e-05,
+      "loss": 2.5635,
+      "step": 9298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.870961769878738e-05,
+      "loss": 0.0034,
+      "step": 9300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8704989354808854e-05,
+      "loss": 3.5664,
+      "step": 9302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.870036101083033e-05,
+      "loss": 0.2119,
+      "step": 9304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8695732666851805e-05,
+      "loss": 1.9195,
+      "step": 9306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8691104322873274e-05,
+      "loss": 1.6738,
+      "step": 9308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.868647597889475e-05,
+      "loss": 6.7344,
+      "step": 9310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8681847634916225e-05,
+      "loss": 0.2258,
+      "step": 9312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.86772192909377e-05,
+      "loss": 1.8337,
+      "step": 9314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8672590946959176e-05,
+      "loss": 3.2966,
+      "step": 9316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.866796260298065e-05,
+      "loss": 6.8161,
+      "step": 9318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8663334259002126e-05,
+      "loss": 1.1347,
+      "step": 9320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8658705915023605e-05,
+      "loss": 4.8397,
+      "step": 9322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.865407757104508e-05,
+      "loss": 0.7547,
+      "step": 9324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8649449227066556e-05,
+      "loss": 1.6403,
+      "step": 9326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.864482088308803e-05,
+      "loss": 1.6045,
+      "step": 9328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8640192539109507e-05,
+      "loss": 1.1348,
+      "step": 9330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8635564195130982e-05,
+      "loss": 2.474,
+      "step": 9332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8630935851152458e-05,
+      "loss": 1.4309,
+      "step": 9334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8626307507173933e-05,
+      "loss": 2.5192,
+      "step": 9336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.862167916319541e-05,
+      "loss": 1.2003,
+      "step": 9338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8617050819216884e-05,
+      "loss": 0.2768,
+      "step": 9340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.861242247523836e-05,
+      "loss": 2.0428,
+      "step": 9342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8607794131259835e-05,
+      "loss": 4.3979,
+      "step": 9344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.860316578728131e-05,
+      "loss": 0.0649,
+      "step": 9346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8598537443302786e-05,
+      "loss": 3.661,
+      "step": 9348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.859390909932426e-05,
+      "loss": 0.3039,
+      "step": 9350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8589280755345737e-05,
+      "loss": 0.3838,
+      "step": 9352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8584652411367212e-05,
+      "loss": 0.7316,
+      "step": 9354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8580024067388687e-05,
+      "loss": 6.011,
+      "step": 9356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8575395723410163e-05,
+      "loss": 3.3309,
+      "step": 9358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8570767379431638e-05,
+      "loss": 2.4937,
+      "step": 9360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8566139035453117e-05,
+      "loss": 0.0009,
+      "step": 9362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8561510691474593e-05,
+      "loss": 1.7528,
+      "step": 9364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8556882347496068e-05,
+      "loss": 4.3861,
+      "step": 9366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8552254003517543e-05,
+      "loss": 0.9031,
+      "step": 9368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.854762565953902e-05,
+      "loss": 2.1664,
+      "step": 9370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8542997315560494e-05,
+      "loss": 2.4013,
+      "step": 9372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.853836897158197e-05,
+      "loss": 7.9108,
+      "step": 9374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8533740627603445e-05,
+      "loss": 5.1065,
+      "step": 9376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.852911228362492e-05,
+      "loss": 0.0186,
+      "step": 9378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8524483939646396e-05,
+      "loss": 0.1255,
+      "step": 9380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.851985559566787e-05,
+      "loss": 7.1159,
+      "step": 9382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8515227251689347e-05,
+      "loss": 2.1793,
+      "step": 9384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8510598907710822e-05,
+      "loss": 2.1917,
+      "step": 9386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8505970563732298e-05,
+      "loss": 0.5336,
+      "step": 9388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8501342219753773e-05,
+      "loss": 0.0655,
+      "step": 9390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.849671387577525e-05,
+      "loss": 1.3722,
+      "step": 9392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8492085531796724e-05,
+      "loss": 2.5299,
+      "step": 9394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.84874571878182e-05,
+      "loss": 1.273,
+      "step": 9396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8482828843839675e-05,
+      "loss": 2.2456,
+      "step": 9398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.847820049986115e-05,
+      "loss": 0.2581,
+      "step": 9400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.847357215588263e-05,
+      "loss": 0.0082,
+      "step": 9402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8468943811904104e-05,
+      "loss": 3.5253,
+      "step": 9404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.846431546792558e-05,
+      "loss": 1.4943,
+      "step": 9406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8459687123947055e-05,
+      "loss": 1.9791,
+      "step": 9408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.845505877996853e-05,
+      "loss": 0.0008,
+      "step": 9410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8450430435990006e-05,
+      "loss": 0.1347,
+      "step": 9412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.844580209201148e-05,
+      "loss": 0.0006,
+      "step": 9414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8441173748032957e-05,
+      "loss": 0.1223,
+      "step": 9416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8436545404054432e-05,
+      "loss": 1.8343,
+      "step": 9418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8431917060075908e-05,
+      "loss": 0.4689,
+      "step": 9420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8427288716097383e-05,
+      "loss": 2.1567,
+      "step": 9422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.842266037211886e-05,
+      "loss": 3.5205,
+      "step": 9424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8418032028140334e-05,
+      "loss": 6.0939,
+      "step": 9426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.841340368416181e-05,
+      "loss": 3.9867,
+      "step": 9428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8408775340183285e-05,
+      "loss": 1.5591,
+      "step": 9430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.840414699620476e-05,
+      "loss": 0.8507,
+      "step": 9432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8399518652226236e-05,
+      "loss": 2.7592,
+      "step": 9434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.839489030824771e-05,
+      "loss": 4.1077,
+      "step": 9436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8390261964269187e-05,
+      "loss": 2.3314,
+      "step": 9438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8385633620290662e-05,
+      "loss": 2.6472,
+      "step": 9440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.838100527631214e-05,
+      "loss": 1.3439,
+      "step": 9442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8376376932333616e-05,
+      "loss": 0.9454,
+      "step": 9444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8371748588355092e-05,
+      "loss": 4.5745,
+      "step": 9446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8367120244376567e-05,
+      "loss": 0.9706,
+      "step": 9448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8362491900398043e-05,
+      "loss": 0.3948,
+      "step": 9450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8357863556419518e-05,
+      "loss": 2.9797,
+      "step": 9452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8353235212440987e-05,
+      "loss": 1.9442,
+      "step": 9454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8348606868462462e-05,
+      "loss": 2.6336,
+      "step": 9456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8343978524483937e-05,
+      "loss": 0.2801,
+      "step": 9458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8339350180505413e-05,
+      "loss": 0.7259,
+      "step": 9460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.833472183652689e-05,
+      "loss": 1.482,
+      "step": 9462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8330093492548364e-05,
+      "loss": 2.2079,
+      "step": 9464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.832546514856984e-05,
+      "loss": 0.3025,
+      "step": 9466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8320836804591318e-05,
+      "loss": 4.4891,
+      "step": 9468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8316208460612793e-05,
+      "loss": 0.0009,
+      "step": 9470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.831158011663427e-05,
+      "loss": 0.3547,
+      "step": 9472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8306951772655744e-05,
+      "loss": 4.8508,
+      "step": 9474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.830232342867722e-05,
+      "loss": 2.8805,
+      "step": 9476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8297695084698695e-05,
+      "loss": 5.0117,
+      "step": 9478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.829306674072017e-05,
+      "loss": 1.8361,
+      "step": 9480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8288438396741646e-05,
+      "loss": 2.4493,
+      "step": 9482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.828381005276312e-05,
+      "loss": 1.9518,
+      "step": 9484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8279181708784597e-05,
+      "loss": 1.2089,
+      "step": 9486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8274553364806072e-05,
+      "loss": 3.7116,
+      "step": 9488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8269925020827548e-05,
+      "loss": 0.9006,
+      "step": 9490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8265296676849023e-05,
+      "loss": 1.6565,
+      "step": 9492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.82606683328705e-05,
+      "loss": 2.5707,
+      "step": 9494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8256039988891974e-05,
+      "loss": 0.2237,
+      "step": 9496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.825141164491345e-05,
+      "loss": 5.3712,
+      "step": 9498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8246783300934925e-05,
+      "loss": 1.2796,
+      "step": 9500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.82421549569564e-05,
+      "loss": 2.1942,
+      "step": 9502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8237526612977876e-05,
+      "loss": 1.2594,
+      "step": 9504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.823289826899935e-05,
+      "loss": 0.3462,
+      "step": 9506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.822826992502083e-05,
+      "loss": 2.4427,
+      "step": 9508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8223641581042305e-05,
+      "loss": 1.4827,
+      "step": 9510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.821901323706378e-05,
+      "loss": 2.0479,
+      "step": 9512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8214384893085256e-05,
+      "loss": 1.7865,
+      "step": 9514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.820975654910673e-05,
+      "loss": 1.7387,
+      "step": 9516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 2.3466,
+      "step": 9518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8200499861149682e-05,
+      "loss": 0.1736,
+      "step": 9520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8195871517171158e-05,
+      "loss": 1.1938,
+      "step": 9522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8191243173192633e-05,
+      "loss": 0.1916,
+      "step": 9524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.818661482921411e-05,
+      "loss": 1.0319,
+      "step": 9526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8181986485235584e-05,
+      "loss": 2.2709,
+      "step": 9528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.817735814125706e-05,
+      "loss": 2.6727,
+      "step": 9530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8172729797278535e-05,
+      "loss": 3.49,
+      "step": 9532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.816810145330001e-05,
+      "loss": 3.937,
+      "step": 9534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8163473109321486e-05,
+      "loss": 8.0554,
+      "step": 9536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.815884476534296e-05,
+      "loss": 2.6918,
+      "step": 9538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8154216421364437e-05,
+      "loss": 3.1106,
+      "step": 9540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8149588077385912e-05,
+      "loss": 0.8753,
+      "step": 9542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8144959733407388e-05,
+      "loss": 0.0437,
+      "step": 9544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8140331389428863e-05,
+      "loss": 2.1921,
+      "step": 9546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8135703045450342e-05,
+      "loss": 2.6359,
+      "step": 9548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8131074701471817e-05,
+      "loss": 4.0934,
+      "step": 9550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8126446357493293e-05,
+      "loss": 3.782,
+      "step": 9552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8121818013514768e-05,
+      "loss": 2.957,
+      "step": 9554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8117189669536243e-05,
+      "loss": 0.0753,
+      "step": 9556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.811256132555772e-05,
+      "loss": 0.677,
+      "step": 9558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8107932981579194e-05,
+      "loss": 1.9879,
+      "step": 9560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.810330463760067e-05,
+      "loss": 2.704,
+      "step": 9562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8098676293622145e-05,
+      "loss": 2.9169,
+      "step": 9564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.809404794964362e-05,
+      "loss": 4.9003,
+      "step": 9566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8089419605665096e-05,
+      "loss": 0.9307,
+      "step": 9568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.808479126168657e-05,
+      "loss": 0.5483,
+      "step": 9570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8080162917708047e-05,
+      "loss": 4.3179,
+      "step": 9572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8075534573729522e-05,
+      "loss": 4.3144,
+      "step": 9574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8070906229750998e-05,
+      "loss": 0.04,
+      "step": 9576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8066277885772473e-05,
+      "loss": 0.9758,
+      "step": 9578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.806164954179395e-05,
+      "loss": 3.2048,
+      "step": 9580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8057021197815424e-05,
+      "loss": 3.7609,
+      "step": 9582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.80523928538369e-05,
+      "loss": 3.0001,
+      "step": 9584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8047764509858375e-05,
+      "loss": 0.3774,
+      "step": 9586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.804313616587985e-05,
+      "loss": 3.8379,
+      "step": 9588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.803850782190133e-05,
+      "loss": 3.1955,
+      "step": 9590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8033879477922804e-05,
+      "loss": 2.2981,
+      "step": 9592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.802925113394428e-05,
+      "loss": 0.4182,
+      "step": 9594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8024622789965755e-05,
+      "loss": 0.5038,
+      "step": 9596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8019994445987224e-05,
+      "loss": 2.1631,
+      "step": 9598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.80153661020087e-05,
+      "loss": 0.005,
+      "step": 9600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8010737758030175e-05,
+      "loss": 0.27,
+      "step": 9602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.800610941405165e-05,
+      "loss": 0.4997,
+      "step": 9604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8001481070073126e-05,
+      "loss": 0.849,
+      "step": 9606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.79968527260946e-05,
+      "loss": 0.0005,
+      "step": 9608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7992224382116077e-05,
+      "loss": 3.8108,
+      "step": 9610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7987596038137552e-05,
+      "loss": 0.3782,
+      "step": 9612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.798296769415903e-05,
+      "loss": 3.6155,
+      "step": 9614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7978339350180506e-05,
+      "loss": 1.7357,
+      "step": 9616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.797371100620198e-05,
+      "loss": 4.0513,
+      "step": 9618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7969082662223457e-05,
+      "loss": 0.3143,
+      "step": 9620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7964454318244932e-05,
+      "loss": 3.141,
+      "step": 9622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7959825974266408e-05,
+      "loss": 2.8876,
+      "step": 9624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7955197630287883e-05,
+      "loss": 3.5034,
+      "step": 9626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.795056928630936e-05,
+      "loss": 1.4211,
+      "step": 9628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7945940942330834e-05,
+      "loss": 0.4753,
+      "step": 9630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.794131259835231e-05,
+      "loss": 1.6033,
+      "step": 9632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7936684254373785e-05,
+      "loss": 0.1687,
+      "step": 9634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.793205591039526e-05,
+      "loss": 0.2104,
+      "step": 9636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7927427566416736e-05,
+      "loss": 2.9397,
+      "step": 9638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.792279922243821e-05,
+      "loss": 2.9196,
+      "step": 9640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7918170878459687e-05,
+      "loss": 0.4265,
+      "step": 9642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7913542534481162e-05,
+      "loss": 1.0265,
+      "step": 9644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7908914190502638e-05,
+      "loss": 5.5529,
+      "step": 9646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7904285846524113e-05,
+      "loss": 0.1232,
+      "step": 9648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.789965750254559e-05,
+      "loss": 4.4074,
+      "step": 9650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7895029158567064e-05,
+      "loss": 0.2816,
+      "step": 9652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.789040081458854e-05,
+      "loss": 0.0003,
+      "step": 9654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7885772470610018e-05,
+      "loss": 0.0012,
+      "step": 9656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7881144126631493e-05,
+      "loss": 3.6426,
+      "step": 9658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.787651578265297e-05,
+      "loss": 4.7657,
+      "step": 9660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7871887438674444e-05,
+      "loss": 0.0009,
+      "step": 9662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.786725909469592e-05,
+      "loss": 4.2075,
+      "step": 9664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7862630750717395e-05,
+      "loss": 1.7408,
+      "step": 9666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.785800240673887e-05,
+      "loss": 2.6668,
+      "step": 9668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7853374062760346e-05,
+      "loss": 2.3333,
+      "step": 9670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.784874571878182e-05,
+      "loss": 2.598,
+      "step": 9672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7844117374803297e-05,
+      "loss": 1.7312,
+      "step": 9674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7839489030824772e-05,
+      "loss": 0.1376,
+      "step": 9676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7834860686846248e-05,
+      "loss": 3.4575,
+      "step": 9678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7830232342867723e-05,
+      "loss": 2.3636,
+      "step": 9680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.78256039988892e-05,
+      "loss": 3.7814,
+      "step": 9682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7820975654910674e-05,
+      "loss": 3.0637,
+      "step": 9684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.781634731093215e-05,
+      "loss": 1.0514,
+      "step": 9686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7811718966953625e-05,
+      "loss": 1.4827,
+      "step": 9688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.78070906229751e-05,
+      "loss": 3.6337,
+      "step": 9690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7802462278996576e-05,
+      "loss": 0.7528,
+      "step": 9692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.779783393501805e-05,
+      "loss": 1.1515,
+      "step": 9694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.779320559103953e-05,
+      "loss": 0.2627,
+      "step": 9696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7788577247061005e-05,
+      "loss": 2.9235,
+      "step": 9698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.778394890308248e-05,
+      "loss": 4.729,
+      "step": 9700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7779320559103956e-05,
+      "loss": 3.2889,
+      "step": 9702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.777469221512543e-05,
+      "loss": 5.1229,
+      "step": 9704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7770063871146907e-05,
+      "loss": 0.458,
+      "step": 9706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7765435527168382e-05,
+      "loss": 0.0043,
+      "step": 9708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7760807183189858e-05,
+      "loss": 1.3344,
+      "step": 9710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7756178839211333e-05,
+      "loss": 3.7577,
+      "step": 9712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.775155049523281e-05,
+      "loss": 4.0129,
+      "step": 9714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7746922151254284e-05,
+      "loss": 0.0063,
+      "step": 9716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.774229380727576e-05,
+      "loss": 0.8528,
+      "step": 9718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7737665463297235e-05,
+      "loss": 1.8749,
+      "step": 9720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.773303711931871e-05,
+      "loss": 3.8136,
+      "step": 9722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7728408775340186e-05,
+      "loss": 1.4091,
+      "step": 9724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.772378043136166e-05,
+      "loss": 5.1273,
+      "step": 9726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7719152087383137e-05,
+      "loss": 5.8095,
+      "step": 9728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7714523743404612e-05,
+      "loss": 6.819,
+      "step": 9730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7709895399426088e-05,
+      "loss": 0.5044,
+      "step": 9732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7705267055447563e-05,
+      "loss": 3.8536,
+      "step": 9734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7700638711469042e-05,
+      "loss": 3.6736,
+      "step": 9736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7696010367490517e-05,
+      "loss": 0.2552,
+      "step": 9738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7691382023511993e-05,
+      "loss": 0.9061,
+      "step": 9740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7686753679533468e-05,
+      "loss": 0.11,
+      "step": 9742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7682125335554937e-05,
+      "loss": 3.1121,
+      "step": 9744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7677496991576412e-05,
+      "loss": 0.852,
+      "step": 9746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7672868647597888e-05,
+      "loss": 1.3867,
+      "step": 9748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7668240303619363e-05,
+      "loss": 3.2169,
+      "step": 9750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.766361195964084e-05,
+      "loss": 3.408,
+      "step": 9752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7658983615662314e-05,
+      "loss": 0.0163,
+      "step": 9754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.765435527168379e-05,
+      "loss": 3.2228,
+      "step": 9756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7649726927705265e-05,
+      "loss": 0.6103,
+      "step": 9758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.764509858372674e-05,
+      "loss": 3.4852,
+      "step": 9760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.764047023974822e-05,
+      "loss": 3.7587,
+      "step": 9762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7635841895769694e-05,
+      "loss": 2.9798,
+      "step": 9764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.763121355179117e-05,
+      "loss": 2.3992,
+      "step": 9766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7626585207812645e-05,
+      "loss": 0.5459,
+      "step": 9768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.762195686383412e-05,
+      "loss": 4.9644,
+      "step": 9770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7617328519855596e-05,
+      "loss": 2.9751,
+      "step": 9772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.761270017587707e-05,
+      "loss": 0.9456,
+      "step": 9774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7608071831898547e-05,
+      "loss": 0.6434,
+      "step": 9776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7603443487920022e-05,
+      "loss": 1.9834,
+      "step": 9778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7598815143941498e-05,
+      "loss": 2.2792,
+      "step": 9780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7594186799962973e-05,
+      "loss": 1.9386,
+      "step": 9782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.758955845598445e-05,
+      "loss": 1.0286,
+      "step": 9784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7584930112005924e-05,
+      "loss": 1.9593,
+      "step": 9786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.75803017680274e-05,
+      "loss": 0.4412,
+      "step": 9788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7575673424048875e-05,
+      "loss": 3.3144,
+      "step": 9790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.757104508007035e-05,
+      "loss": 1.1425,
+      "step": 9792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7566416736091826e-05,
+      "loss": 2.6171,
+      "step": 9794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.75617883921133e-05,
+      "loss": 0.0107,
+      "step": 9796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7557160048134777e-05,
+      "loss": 1.4387,
+      "step": 9798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7552531704156252e-05,
+      "loss": 0.474,
+      "step": 9800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.754790336017773e-05,
+      "loss": 1.4048,
+      "step": 9802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7543275016199206e-05,
+      "loss": 5.2465,
+      "step": 9804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.753864667222068e-05,
+      "loss": 1.7191,
+      "step": 9806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7534018328242157e-05,
+      "loss": 1.6944,
+      "step": 9808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7529389984263633e-05,
+      "loss": 0.2251,
+      "step": 9810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7524761640285108e-05,
+      "loss": 1.6832,
+      "step": 9812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7520133296306583e-05,
+      "loss": 1.275,
+      "step": 9814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.751550495232806e-05,
+      "loss": 2.7259,
+      "step": 9816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7510876608349534e-05,
+      "loss": 3.7165,
+      "step": 9818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.750624826437101e-05,
+      "loss": 0.3067,
+      "step": 9820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7501619920392485e-05,
+      "loss": 0.553,
+      "step": 9822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.749699157641396e-05,
+      "loss": 1.184,
+      "step": 9824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7492363232435436e-05,
+      "loss": 4.8073,
+      "step": 9826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.748773488845691e-05,
+      "loss": 1.6833,
+      "step": 9828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7483106544478387e-05,
+      "loss": 0.0097,
+      "step": 9830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7478478200499862e-05,
+      "loss": 0.9655,
+      "step": 9832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7473849856521338e-05,
+      "loss": 1.0422,
+      "step": 9834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7469221512542813e-05,
+      "loss": 2.4773,
+      "step": 9836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.746459316856429e-05,
+      "loss": 2.4311,
+      "step": 9838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7459964824585764e-05,
+      "loss": 0.0034,
+      "step": 9840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7455336480607243e-05,
+      "loss": 4.1189,
+      "step": 9842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7450708136628718e-05,
+      "loss": 3.2575,
+      "step": 9844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7446079792650194e-05,
+      "loss": 3.8947,
+      "step": 9846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.744145144867167e-05,
+      "loss": 1.9695,
+      "step": 9848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7436823104693144e-05,
+      "loss": 3.65,
+      "step": 9850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.743219476071462e-05,
+      "loss": 0.4789,
+      "step": 9852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7427566416736095e-05,
+      "loss": 0.2113,
+      "step": 9854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.742293807275757e-05,
+      "loss": 4.6576,
+      "step": 9856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7418309728779046e-05,
+      "loss": 0.9281,
+      "step": 9858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.741368138480052e-05,
+      "loss": 4.1118,
+      "step": 9860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7409053040821997e-05,
+      "loss": 1.7478,
+      "step": 9862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7404424696843472e-05,
+      "loss": 0.0012,
+      "step": 9864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7399796352864948e-05,
+      "loss": 2.7279,
+      "step": 9866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7395168008886423e-05,
+      "loss": 2.7854,
+      "step": 9868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.73905396649079e-05,
+      "loss": 1.2178,
+      "step": 9870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7385911320929374e-05,
+      "loss": 4.0006,
+      "step": 9872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.738128297695085e-05,
+      "loss": 1.7411,
+      "step": 9874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7376654632972325e-05,
+      "loss": 1.462,
+      "step": 9876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.73720262889938e-05,
+      "loss": 0.6637,
+      "step": 9878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7367397945015276e-05,
+      "loss": 0.5077,
+      "step": 9880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7362769601036755e-05,
+      "loss": 1.229,
+      "step": 9882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.735814125705823e-05,
+      "loss": 3.3321,
+      "step": 9884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7353512913079705e-05,
+      "loss": 0.1546,
+      "step": 9886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7348884569101174e-05,
+      "loss": 2.597,
+      "step": 9888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.734425622512265e-05,
+      "loss": 5.6346,
+      "step": 9890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7339627881144125e-05,
+      "loss": 0.887,
+      "step": 9892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.73349995371656e-05,
+      "loss": 0.0004,
+      "step": 9894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7330371193187076e-05,
+      "loss": 1.9226,
+      "step": 9896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.732574284920855e-05,
+      "loss": 2.2585,
+      "step": 9898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7321114505230027e-05,
+      "loss": 4.8726,
+      "step": 9900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7316486161251502e-05,
+      "loss": 8.3477,
+      "step": 9902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7311857817272977e-05,
+      "loss": 0.0015,
+      "step": 9904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7307229473294453e-05,
+      "loss": 1.7323,
+      "step": 9906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7302601129315932e-05,
+      "loss": 2.5225,
+      "step": 9908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7297972785337407e-05,
+      "loss": 3.0766,
+      "step": 9910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7293344441358883e-05,
+      "loss": 1.5634,
+      "step": 9912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7288716097380358e-05,
+      "loss": 1.568,
+      "step": 9914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7284087753401833e-05,
+      "loss": 0.2167,
+      "step": 9916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.727945940942331e-05,
+      "loss": 2.3831,
+      "step": 9918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7274831065444784e-05,
+      "loss": 0.6455,
+      "step": 9920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.727020272146626e-05,
+      "loss": 1.2424,
+      "step": 9922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7265574377487735e-05,
+      "loss": 3.6001,
+      "step": 9924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.726094603350921e-05,
+      "loss": 1.0683,
+      "step": 9926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7256317689530686e-05,
+      "loss": 2.2093,
+      "step": 9928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.725168934555216e-05,
+      "loss": 4.8441,
+      "step": 9930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7247061001573637e-05,
+      "loss": 3.5826,
+      "step": 9932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7242432657595112e-05,
+      "loss": 0.544,
+      "step": 9934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7237804313616588e-05,
+      "loss": 1.5095,
+      "step": 9936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7233175969638063e-05,
+      "loss": 1.7371,
+      "step": 9938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.722854762565954e-05,
+      "loss": 5.2835,
+      "step": 9940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7223919281681014e-05,
+      "loss": 1.1688,
+      "step": 9942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.721929093770249e-05,
+      "loss": 6.0212,
+      "step": 9944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7214662593723965e-05,
+      "loss": 3.1075,
+      "step": 9946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.721003424974544e-05,
+      "loss": 5.3276,
+      "step": 9948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.720540590576692e-05,
+      "loss": 0.0154,
+      "step": 9950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7200777561788394e-05,
+      "loss": 2.2289,
+      "step": 9952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.719614921780987e-05,
+      "loss": 1.07,
+      "step": 9954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7191520873831345e-05,
+      "loss": 0.3311,
+      "step": 9956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.718689252985282e-05,
+      "loss": 3.1961,
+      "step": 9958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7182264185874296e-05,
+      "loss": 0.5719,
+      "step": 9960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.717763584189577e-05,
+      "loss": 2.1008,
+      "step": 9962
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7173007497917247e-05,
+      "loss": 6.252,
+      "step": 9964
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7168379153938722e-05,
+      "loss": 3.3756,
+      "step": 9966
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7163750809960198e-05,
+      "loss": 0.6846,
+      "step": 9968
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7159122465981673e-05,
+      "loss": 3.6695,
+      "step": 9970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.715449412200315e-05,
+      "loss": 2.0568,
+      "step": 9972
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7149865778024624e-05,
+      "loss": 0.5983,
+      "step": 9974
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.71452374340461e-05,
+      "loss": 1.7116,
+      "step": 9976
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7140609090067575e-05,
+      "loss": 2.5836,
+      "step": 9978
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.713598074608905e-05,
+      "loss": 1.2019,
+      "step": 9980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7131352402110526e-05,
+      "loss": 0.7982,
+      "step": 9982
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7126724058132e-05,
+      "loss": 2.6695,
+      "step": 9984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7122095714153477e-05,
+      "loss": 1.2676,
+      "step": 9986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7117467370174952e-05,
+      "loss": 0.1504,
+      "step": 9988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.711283902619643e-05,
+      "loss": 4.8052,
+      "step": 9990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7108210682217906e-05,
+      "loss": 2.5771,
+      "step": 9992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7103582338239382e-05,
+      "loss": 0.0068,
+      "step": 9994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7098953994260857e-05,
+      "loss": 1.5693,
+      "step": 9996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7094325650282333e-05,
+      "loss": 2.9367,
+      "step": 9998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7089697306303808e-05,
+      "loss": 5.1741,
+      "step": 10000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7085068962325283e-05,
+      "loss": 2.7252,
+      "step": 10002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.708044061834676e-05,
+      "loss": 3.5361,
+      "step": 10004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7075812274368234e-05,
+      "loss": 1.2624,
+      "step": 10006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.707118393038971e-05,
+      "loss": 3.4286,
+      "step": 10008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7066555586411185e-05,
+      "loss": 1.152,
+      "step": 10010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.706192724243266e-05,
+      "loss": 0.0743,
+      "step": 10012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7057298898454136e-05,
+      "loss": 1.5451,
+      "step": 10014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.705267055447561e-05,
+      "loss": 0.0262,
+      "step": 10016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7048042210497087e-05,
+      "loss": 2.2949,
+      "step": 10018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7043413866518562e-05,
+      "loss": 5.0104,
+      "step": 10020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7038785522540038e-05,
+      "loss": 1.6818,
+      "step": 10022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7034157178561513e-05,
+      "loss": 0.0146,
+      "step": 10024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.702952883458299e-05,
+      "loss": 4.5547,
+      "step": 10026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7024900490604464e-05,
+      "loss": 1.7554,
+      "step": 10028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7020272146625943e-05,
+      "loss": 1.0651,
+      "step": 10030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7015643802647418e-05,
+      "loss": 2.2288,
+      "step": 10032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7011015458668887e-05,
+      "loss": 1.3415,
+      "step": 10034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7006387114690362e-05,
+      "loss": 3.5655,
+      "step": 10036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.7001758770711838e-05,
+      "loss": 0.9472,
+      "step": 10038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6997130426733313e-05,
+      "loss": 2.5263,
+      "step": 10040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.699250208275479e-05,
+      "loss": 2.1224,
+      "step": 10042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6987873738776264e-05,
+      "loss": 0.3059,
+      "step": 10044
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.698324539479774e-05,
+      "loss": 0.0036,
+      "step": 10046
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6978617050819215e-05,
+      "loss": 0.1746,
+      "step": 10048
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.697398870684069e-05,
+      "loss": 1.8547,
+      "step": 10050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6969360362862166e-05,
+      "loss": 0.7831,
+      "step": 10052
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.696473201888364e-05,
+      "loss": 4.4959,
+      "step": 10054
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.696010367490512e-05,
+      "loss": 0.1684,
+      "step": 10056
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6955475330926595e-05,
+      "loss": 1.3188,
+      "step": 10058
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.695084698694807e-05,
+      "loss": 2.7311,
+      "step": 10060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6946218642969546e-05,
+      "loss": 3.0088,
+      "step": 10062
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.694159029899102e-05,
+      "loss": 2.102,
+      "step": 10064
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6936961955012497e-05,
+      "loss": 1.3942,
+      "step": 10066
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6932333611033972e-05,
+      "loss": 1.4054,
+      "step": 10068
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6927705267055448e-05,
+      "loss": 1.7755,
+      "step": 10070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6923076923076923e-05,
+      "loss": 0.9476,
+      "step": 10072
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.69184485790984e-05,
+      "loss": 0.2252,
+      "step": 10074
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6913820235119874e-05,
+      "loss": 2.3343,
+      "step": 10076
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.690919189114135e-05,
+      "loss": 2.2354,
+      "step": 10078
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6904563547162825e-05,
+      "loss": 1.7886,
+      "step": 10080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.68999352031843e-05,
+      "loss": 1.6227,
+      "step": 10082
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6895306859205776e-05,
+      "loss": 2.278,
+      "step": 10084
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.689067851522725e-05,
+      "loss": 1.4469,
+      "step": 10086
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6886050171248727e-05,
+      "loss": 2.5364,
+      "step": 10088
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6881421827270202e-05,
+      "loss": 0.8596,
+      "step": 10090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6876793483291678e-05,
+      "loss": 1.5537,
+      "step": 10092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6872165139313153e-05,
+      "loss": 1.575,
+      "step": 10094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6867536795334632e-05,
+      "loss": 1.0405,
+      "step": 10096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6862908451356107e-05,
+      "loss": 2.3742,
+      "step": 10098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6858280107377583e-05,
+      "loss": 2.3216,
+      "step": 10100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6853651763399058e-05,
+      "loss": 2.5607,
+      "step": 10102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6849023419420534e-05,
+      "loss": 1.0264,
+      "step": 10104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.684439507544201e-05,
+      "loss": 0.7691,
+      "step": 10106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6839766731463484e-05,
+      "loss": 0.9071,
+      "step": 10108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.683513838748496e-05,
+      "loss": 4.424,
+      "step": 10110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6830510043506435e-05,
+      "loss": 0.4335,
+      "step": 10112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.682588169952791e-05,
+      "loss": 0.2197,
+      "step": 10114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6821253355549386e-05,
+      "loss": 2.2308,
+      "step": 10116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.681662501157086e-05,
+      "loss": 2.1068,
+      "step": 10118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6811996667592337e-05,
+      "loss": 4.6484,
+      "step": 10120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6807368323613812e-05,
+      "loss": 4.2568,
+      "step": 10122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6802739979635288e-05,
+      "loss": 4.9319,
+      "step": 10124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6798111635656763e-05,
+      "loss": 9.7115,
+      "step": 10126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.679348329167824e-05,
+      "loss": 0.0059,
+      "step": 10128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6788854947699714e-05,
+      "loss": 1.317,
+      "step": 10130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.678422660372119e-05,
+      "loss": 6.1077,
+      "step": 10132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6779598259742665e-05,
+      "loss": 4.0945,
+      "step": 10134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6774969915764144e-05,
+      "loss": 0.5854,
+      "step": 10136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.677034157178562e-05,
+      "loss": 1.2135,
+      "step": 10138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6765713227807095e-05,
+      "loss": 0.8805,
+      "step": 10140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.676108488382857e-05,
+      "loss": 1.7084,
+      "step": 10142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6756456539850045e-05,
+      "loss": 0.8596,
+      "step": 10144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.675182819587152e-05,
+      "loss": 0.0085,
+      "step": 10146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6747199851892996e-05,
+      "loss": 1.4268,
+      "step": 10148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.674257150791447e-05,
+      "loss": 0.2733,
+      "step": 10150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6737943163935947e-05,
+      "loss": 7.3822,
+      "step": 10152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6733314819957423e-05,
+      "loss": 6.5193,
+      "step": 10154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6728686475978898e-05,
+      "loss": 0.9149,
+      "step": 10156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6724058132000373e-05,
+      "loss": 5.5151,
+      "step": 10158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.671942978802185e-05,
+      "loss": 2.9135,
+      "step": 10160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6714801444043324e-05,
+      "loss": 1.7214,
+      "step": 10162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.67101731000648e-05,
+      "loss": 0.4486,
+      "step": 10164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6705544756086275e-05,
+      "loss": 0.1564,
+      "step": 10166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.670091641210775e-05,
+      "loss": 0.6106,
+      "step": 10168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6696288068129226e-05,
+      "loss": 1.3377,
+      "step": 10170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.66916597241507e-05,
+      "loss": 1.9144,
+      "step": 10172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6687031380172177e-05,
+      "loss": 0.0046,
+      "step": 10174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6682403036193656e-05,
+      "loss": 0.8383,
+      "step": 10176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6677774692215124e-05,
+      "loss": 1.1476,
+      "step": 10178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.66731463482366e-05,
+      "loss": 3.8101,
+      "step": 10180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6668518004258075e-05,
+      "loss": 4.6926,
+      "step": 10182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.666388966027955e-05,
+      "loss": 1.6734,
+      "step": 10184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6659261316301026e-05,
+      "loss": 0.1078,
+      "step": 10186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.66546329723225e-05,
+      "loss": 0.0044,
+      "step": 10188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6650004628343977e-05,
+      "loss": 2.0899,
+      "step": 10190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6645376284365452e-05,
+      "loss": 0.061,
+      "step": 10192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6640747940386928e-05,
+      "loss": 5.3708,
+      "step": 10194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6636119596408403e-05,
+      "loss": 0.4772,
+      "step": 10196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.663149125242988e-05,
+      "loss": 0.3239,
+      "step": 10198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6626862908451354e-05,
+      "loss": 1.8412,
+      "step": 10200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6622234564472833e-05,
+      "loss": 0.2176,
+      "step": 10202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6617606220494308e-05,
+      "loss": 0.0006,
+      "step": 10204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6612977876515784e-05,
+      "loss": 0.0023,
+      "step": 10206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.660834953253726e-05,
+      "loss": 0.3058,
+      "step": 10208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6603721188558734e-05,
+      "loss": 2.0413,
+      "step": 10210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.659909284458021e-05,
+      "loss": 6.2956,
+      "step": 10212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6594464500601685e-05,
+      "loss": 1.2816,
+      "step": 10214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.658983615662316e-05,
+      "loss": 1.4848,
+      "step": 10216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6585207812644636e-05,
+      "loss": 0.0694,
+      "step": 10218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.658057946866611e-05,
+      "loss": 6.6251,
+      "step": 10220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6575951124687587e-05,
+      "loss": 0.1252,
+      "step": 10222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6571322780709062e-05,
+      "loss": 0.9605,
+      "step": 10224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6566694436730538e-05,
+      "loss": 0.0045,
+      "step": 10226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6562066092752013e-05,
+      "loss": 8.2282,
+      "step": 10228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.655743774877349e-05,
+      "loss": 4.4762,
+      "step": 10230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6552809404794964e-05,
+      "loss": 0.3105,
+      "step": 10232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.654818106081644e-05,
+      "loss": 0.0005,
+      "step": 10234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6543552716837915e-05,
+      "loss": 0.135,
+      "step": 10236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.653892437285939e-05,
+      "loss": 0.3372,
+      "step": 10238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6534296028880866e-05,
+      "loss": 8.881,
+      "step": 10240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6529667684902345e-05,
+      "loss": 3.5332,
+      "step": 10242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.652503934092382e-05,
+      "loss": 4.2912,
+      "step": 10244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6520410996945295e-05,
+      "loss": 0.5557,
+      "step": 10246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.651578265296677e-05,
+      "loss": 2.1343,
+      "step": 10248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6511154308988246e-05,
+      "loss": 4.5437,
+      "step": 10250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6506525965009722e-05,
+      "loss": 7.403,
+      "step": 10252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6501897621031197e-05,
+      "loss": 0.3731,
+      "step": 10254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6497269277052673e-05,
+      "loss": 0.4076,
+      "step": 10256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6492640933074148e-05,
+      "loss": 0.5089,
+      "step": 10258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6488012589095623e-05,
+      "loss": 1.3773,
+      "step": 10260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.64833842451171e-05,
+      "loss": 0.8359,
+      "step": 10262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6478755901138574e-05,
+      "loss": 0.0019,
+      "step": 10264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.647412755716005e-05,
+      "loss": 2.3594,
+      "step": 10266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6469499213181525e-05,
+      "loss": 2.0376,
+      "step": 10268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6464870869203e-05,
+      "loss": 0.1366,
+      "step": 10270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6460242525224476e-05,
+      "loss": 4.2354,
+      "step": 10272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.645561418124595e-05,
+      "loss": 5.1587,
+      "step": 10274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6450985837267427e-05,
+      "loss": 3.3224,
+      "step": 10276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6446357493288902e-05,
+      "loss": 0.055,
+      "step": 10278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6441729149310378e-05,
+      "loss": 0.0662,
+      "step": 10280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6437100805331853e-05,
+      "loss": 3.1553,
+      "step": 10282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6432472461353332e-05,
+      "loss": 2.8004,
+      "step": 10284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6427844117374807e-05,
+      "loss": 1.2614,
+      "step": 10286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6423215773396283e-05,
+      "loss": 2.2809,
+      "step": 10288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6418587429417758e-05,
+      "loss": 2.2818,
+      "step": 10290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6413959085439234e-05,
+      "loss": 1.3169,
+      "step": 10292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.640933074146071e-05,
+      "loss": 2.6348,
+      "step": 10294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6404702397482184e-05,
+      "loss": 1.548,
+      "step": 10296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.640007405350366e-05,
+      "loss": 6.3316,
+      "step": 10298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6395445709525135e-05,
+      "loss": 0.0034,
+      "step": 10300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.639081736554661e-05,
+      "loss": 0.0042,
+      "step": 10302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6386189021568086e-05,
+      "loss": 2.9232,
+      "step": 10304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.638156067758956e-05,
+      "loss": 1.1083,
+      "step": 10306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6376932333611037e-05,
+      "loss": 0.0003,
+      "step": 10308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6372303989632512e-05,
+      "loss": 1.7112,
+      "step": 10310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6367675645653988e-05,
+      "loss": 1.4493,
+      "step": 10312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6363047301675463e-05,
+      "loss": 5.4348,
+      "step": 10314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.635841895769694e-05,
+      "loss": 5.6853,
+      "step": 10316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6353790613718414e-05,
+      "loss": 3.2101,
+      "step": 10318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.634916226973989e-05,
+      "loss": 0.0022,
+      "step": 10320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6344533925761365e-05,
+      "loss": 6.6323,
+      "step": 10322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6339905581782837e-05,
+      "loss": 1.1668,
+      "step": 10324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6335277237804312e-05,
+      "loss": 2.2497,
+      "step": 10326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6330648893825788e-05,
+      "loss": 1.3288,
+      "step": 10328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6326020549847263e-05,
+      "loss": 2.636,
+      "step": 10330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.632139220586874e-05,
+      "loss": 3.8662,
+      "step": 10332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6316763861890214e-05,
+      "loss": 0.5036,
+      "step": 10334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.631213551791169e-05,
+      "loss": 3.8454,
+      "step": 10336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6307507173933165e-05,
+      "loss": 0.9342,
+      "step": 10338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.630287882995464e-05,
+      "loss": 0.238,
+      "step": 10340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6298250485976116e-05,
+      "loss": 3.8007,
+      "step": 10342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.629362214199759e-05,
+      "loss": 0.8698,
+      "step": 10344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6288993798019067e-05,
+      "loss": 5.4808,
+      "step": 10346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6284365454040542e-05,
+      "loss": 3.9051,
+      "step": 10348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.627973711006202e-05,
+      "loss": 2.653,
+      "step": 10350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6275108766083496e-05,
+      "loss": 2.5757,
+      "step": 10352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6270480422104972e-05,
+      "loss": 3.8571,
+      "step": 10354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6265852078126447e-05,
+      "loss": 2.7164,
+      "step": 10356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6261223734147923e-05,
+      "loss": 1.8354,
+      "step": 10358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6256595390169398e-05,
+      "loss": 0.7843,
+      "step": 10360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6251967046190873e-05,
+      "loss": 3.1618,
+      "step": 10362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.624733870221235e-05,
+      "loss": 1.8528,
+      "step": 10364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6242710358233824e-05,
+      "loss": 1.1338,
+      "step": 10366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.62380820142553e-05,
+      "loss": 2.4401,
+      "step": 10368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6233453670276775e-05,
+      "loss": 3.8264,
+      "step": 10370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.622882532629825e-05,
+      "loss": 1.6514,
+      "step": 10372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6224196982319726e-05,
+      "loss": 1.358,
+      "step": 10374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.62195686383412e-05,
+      "loss": 1.4559,
+      "step": 10376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6214940294362677e-05,
+      "loss": 1.0523,
+      "step": 10378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6210311950384152e-05,
+      "loss": 0.5671,
+      "step": 10380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6205683606405628e-05,
+      "loss": 1.2218,
+      "step": 10382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6201055262427103e-05,
+      "loss": 0.9899,
+      "step": 10384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.619642691844858e-05,
+      "loss": 0.5512,
+      "step": 10386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6191798574470054e-05,
+      "loss": 0.9641,
+      "step": 10388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6187170230491533e-05,
+      "loss": 1.1174,
+      "step": 10390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6182541886513008e-05,
+      "loss": 0.0414,
+      "step": 10392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6177913542534484e-05,
+      "loss": 1.4995,
+      "step": 10394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.617328519855596e-05,
+      "loss": 5.1025,
+      "step": 10396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6168656854577434e-05,
+      "loss": 0.9862,
+      "step": 10398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.616402851059891e-05,
+      "loss": 1.1396,
+      "step": 10400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6159400166620385e-05,
+      "loss": 2.5759,
+      "step": 10402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.615477182264186e-05,
+      "loss": 1.1888,
+      "step": 10404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6150143478663336e-05,
+      "loss": 0.9193,
+      "step": 10406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.614551513468481e-05,
+      "loss": 0.0948,
+      "step": 10408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6140886790706287e-05,
+      "loss": 0.8643,
+      "step": 10410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6136258446727762e-05,
+      "loss": 6.166,
+      "step": 10412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6131630102749238e-05,
+      "loss": 2.3391,
+      "step": 10414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6127001758770713e-05,
+      "loss": 1.7891,
+      "step": 10416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.612237341479219e-05,
+      "loss": 1.506,
+      "step": 10418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6117745070813664e-05,
+      "loss": 1.6155,
+      "step": 10420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.611311672683514e-05,
+      "loss": 0.9201,
+      "step": 10422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6108488382856615e-05,
+      "loss": 0.017,
+      "step": 10424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.610386003887809e-05,
+      "loss": 0.1159,
+      "step": 10426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6099231694899566e-05,
+      "loss": 2.1721,
+      "step": 10428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6094603350921045e-05,
+      "loss": 1.33,
+      "step": 10430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.608997500694252e-05,
+      "loss": 5.2398,
+      "step": 10432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6085346662963996e-05,
+      "loss": 0.7176,
+      "step": 10434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.608071831898547e-05,
+      "loss": 3.5588,
+      "step": 10436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6076089975006946e-05,
+      "loss": 6.9225,
+      "step": 10438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6071461631028422e-05,
+      "loss": 2.9403,
+      "step": 10440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6066833287049897e-05,
+      "loss": 0.9541,
+      "step": 10442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6062204943071373e-05,
+      "loss": 0.554,
+      "step": 10444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6057576599092848e-05,
+      "loss": 1.3721,
+      "step": 10446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6052948255114323e-05,
+      "loss": 1.1349,
+      "step": 10448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.60483199111358e-05,
+      "loss": 0.9258,
+      "step": 10450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6043691567157274e-05,
+      "loss": 3.2953,
+      "step": 10452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.603906322317875e-05,
+      "loss": 0.8485,
+      "step": 10454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6034434879200225e-05,
+      "loss": 0.2823,
+      "step": 10456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.60298065352217e-05,
+      "loss": 0.8751,
+      "step": 10458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6025178191243176e-05,
+      "loss": 4.899,
+      "step": 10460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.602054984726465e-05,
+      "loss": 0.2176,
+      "step": 10462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6015921503286127e-05,
+      "loss": 1.6826,
+      "step": 10464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6011293159307602e-05,
+      "loss": 0.3138,
+      "step": 10466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6006664815329074e-05,
+      "loss": 2.7041,
+      "step": 10468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.600203647135055e-05,
+      "loss": 6.2093,
+      "step": 10470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5997408127372025e-05,
+      "loss": 0.0022,
+      "step": 10472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.59927797833935e-05,
+      "loss": 2.2286,
+      "step": 10474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5988151439414976e-05,
+      "loss": 6.6755,
+      "step": 10476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.598352309543645e-05,
+      "loss": 4.358,
+      "step": 10478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5978894751457927e-05,
+      "loss": 4.1574,
+      "step": 10480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5974266407479402e-05,
+      "loss": 0.7281,
+      "step": 10482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5969638063500878e-05,
+      "loss": 5.0903,
+      "step": 10484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5965009719522353e-05,
+      "loss": 2.4332,
+      "step": 10486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.596038137554383e-05,
+      "loss": 0.0004,
+      "step": 10488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5955753031565304e-05,
+      "loss": 0.0006,
+      "step": 10490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.595112468758678e-05,
+      "loss": 0.6822,
+      "step": 10492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5946496343608255e-05,
+      "loss": 1.2764,
+      "step": 10494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5941867999629734e-05,
+      "loss": 2.136,
+      "step": 10496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.593723965565121e-05,
+      "loss": 0.0312,
+      "step": 10498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5932611311672685e-05,
+      "loss": 0.4909,
+      "step": 10500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.592798296769416e-05,
+      "loss": 1.482,
+      "step": 10502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5923354623715635e-05,
+      "loss": 1.1747,
+      "step": 10504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.591872627973711e-05,
+      "loss": 3.5803,
+      "step": 10506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5914097935758586e-05,
+      "loss": 3.5772,
+      "step": 10508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.590946959178006e-05,
+      "loss": 1.9199,
+      "step": 10510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5904841247801537e-05,
+      "loss": 0.4333,
+      "step": 10512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5900212903823012e-05,
+      "loss": 2.204,
+      "step": 10514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5895584559844488e-05,
+      "loss": 1.1161,
+      "step": 10516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5890956215865963e-05,
+      "loss": 1.7936,
+      "step": 10518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.588632787188744e-05,
+      "loss": 0.5479,
+      "step": 10520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5881699527908914e-05,
+      "loss": 1.3175,
+      "step": 10522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.587707118393039e-05,
+      "loss": 0.0008,
+      "step": 10524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5872442839951865e-05,
+      "loss": 1.1473,
+      "step": 10526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.586781449597334e-05,
+      "loss": 1.2104,
+      "step": 10528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5863186151994816e-05,
+      "loss": 2.3721,
+      "step": 10530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.585855780801629e-05,
+      "loss": 2.7306,
+      "step": 10532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5853929464037767e-05,
+      "loss": 0.8466,
+      "step": 10534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5849301120059246e-05,
+      "loss": 3.5906,
+      "step": 10536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.584467277608072e-05,
+      "loss": 1.8663,
+      "step": 10538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5840044432102196e-05,
+      "loss": 0.5272,
+      "step": 10540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5835416088123672e-05,
+      "loss": 2.1581,
+      "step": 10542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5830787744145147e-05,
+      "loss": 9.88,
+      "step": 10544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5826159400166623e-05,
+      "loss": 2.718,
+      "step": 10546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5821531056188098e-05,
+      "loss": 5.0854,
+      "step": 10548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5816902712209574e-05,
+      "loss": 3.1532,
+      "step": 10550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.581227436823105e-05,
+      "loss": 0.8753,
+      "step": 10552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5807646024252524e-05,
+      "loss": 0.7775,
+      "step": 10554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5803017680274e-05,
+      "loss": 4.2073,
+      "step": 10556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5798389336295475e-05,
+      "loss": 0.1827,
+      "step": 10558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.579376099231695e-05,
+      "loss": 4.5917,
+      "step": 10560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5789132648338426e-05,
+      "loss": 0.0011,
+      "step": 10562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.57845043043599e-05,
+      "loss": 1.2392,
+      "step": 10564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5779875960381377e-05,
+      "loss": 2.0865,
+      "step": 10566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5775247616402852e-05,
+      "loss": 0.3336,
+      "step": 10568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5770619272424328e-05,
+      "loss": 1.6361,
+      "step": 10570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5765990928445803e-05,
+      "loss": 3.3452,
+      "step": 10572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.576136258446728e-05,
+      "loss": 1.1774,
+      "step": 10574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5756734240488754e-05,
+      "loss": 0.0012,
+      "step": 10576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5752105896510233e-05,
+      "loss": 3.2275,
+      "step": 10578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5747477552531708e-05,
+      "loss": 1.7193,
+      "step": 10580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5742849208553184e-05,
+      "loss": 2.0842,
+      "step": 10582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.573822086457466e-05,
+      "loss": 0.2502,
+      "step": 10584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5733592520596135e-05,
+      "loss": 1.5428,
+      "step": 10586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.572896417661761e-05,
+      "loss": 0.4105,
+      "step": 10588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5724335832639085e-05,
+      "loss": 2.146,
+      "step": 10590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.571970748866056e-05,
+      "loss": 0.2364,
+      "step": 10592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5715079144682036e-05,
+      "loss": 1.5412,
+      "step": 10594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.571045080070351e-05,
+      "loss": 0.9249,
+      "step": 10596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5705822456724987e-05,
+      "loss": 3.7576,
+      "step": 10598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5701194112746463e-05,
+      "loss": 4.3339,
+      "step": 10600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5696565768767938e-05,
+      "loss": 0.6386,
+      "step": 10602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5691937424789413e-05,
+      "loss": 0.2593,
+      "step": 10604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.568730908081089e-05,
+      "loss": 0.0038,
+      "step": 10606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5682680736832364e-05,
+      "loss": 6.044,
+      "step": 10608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.567805239285384e-05,
+      "loss": 0.8263,
+      "step": 10610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5673424048875315e-05,
+      "loss": 3.2659,
+      "step": 10612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5668795704896787e-05,
+      "loss": 2.8776,
+      "step": 10614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5664167360918263e-05,
+      "loss": 4.6936,
+      "step": 10616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5659539016939738e-05,
+      "loss": 1.6824,
+      "step": 10618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5654910672961213e-05,
+      "loss": 2.5735,
+      "step": 10620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.565028232898269e-05,
+      "loss": 4.6955,
+      "step": 10622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5645653985004164e-05,
+      "loss": 1.3968,
+      "step": 10624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 2.956,
+      "step": 10626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5636397297047115e-05,
+      "loss": 0.8258,
+      "step": 10628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.563176895306859e-05,
+      "loss": 2.8366,
+      "step": 10630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5627140609090066e-05,
+      "loss": 0.2113,
+      "step": 10632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.562251226511154e-05,
+      "loss": 1.1126,
+      "step": 10634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5617883921133017e-05,
+      "loss": 1.4459,
+      "step": 10636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5613255577154492e-05,
+      "loss": 0.0003,
+      "step": 10638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5608627233175968e-05,
+      "loss": 1.0867,
+      "step": 10640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5603998889197443e-05,
+      "loss": 3.8417,
+      "step": 10642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5599370545218922e-05,
+      "loss": 0.663,
+      "step": 10644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5594742201240397e-05,
+      "loss": 0.0667,
+      "step": 10646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5590113857261873e-05,
+      "loss": 1.3336,
+      "step": 10648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5585485513283348e-05,
+      "loss": 1.7983,
+      "step": 10650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5580857169304824e-05,
+      "loss": 0.2013,
+      "step": 10652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.55762288253263e-05,
+      "loss": 2.7805,
+      "step": 10654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5571600481347774e-05,
+      "loss": 0.8984,
+      "step": 10656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.556697213736925e-05,
+      "loss": 0.6232,
+      "step": 10658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5562343793390725e-05,
+      "loss": 0.9569,
+      "step": 10660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.55577154494122e-05,
+      "loss": 0.1934,
+      "step": 10662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5553087105433676e-05,
+      "loss": 1.8113,
+      "step": 10664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.554845876145515e-05,
+      "loss": 1.0031,
+      "step": 10666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5543830417476627e-05,
+      "loss": 2.691,
+      "step": 10668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5539202073498102e-05,
+      "loss": 2.0045,
+      "step": 10670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5534573729519578e-05,
+      "loss": 0.1695,
+      "step": 10672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5529945385541053e-05,
+      "loss": 0.9666,
+      "step": 10674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.552531704156253e-05,
+      "loss": 1.0635,
+      "step": 10676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5520688697584004e-05,
+      "loss": 1.9105,
+      "step": 10678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.551606035360548e-05,
+      "loss": 5.6755,
+      "step": 10680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5511432009626955e-05,
+      "loss": 0.7376,
+      "step": 10682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5506803665648434e-05,
+      "loss": 8.4785,
+      "step": 10684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.550217532166991e-05,
+      "loss": 1.5228,
+      "step": 10686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5497546977691385e-05,
+      "loss": 2.5081,
+      "step": 10688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.549291863371286e-05,
+      "loss": 3.8573,
+      "step": 10690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5488290289734335e-05,
+      "loss": 0.9465,
+      "step": 10692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.548366194575581e-05,
+      "loss": 0.3552,
+      "step": 10694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5479033601777286e-05,
+      "loss": 2.617,
+      "step": 10696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5474405257798762e-05,
+      "loss": 2.9995,
+      "step": 10698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5469776913820237e-05,
+      "loss": 1.2429,
+      "step": 10700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5465148569841713e-05,
+      "loss": 1.1822,
+      "step": 10702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5460520225863188e-05,
+      "loss": 0.129,
+      "step": 10704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5455891881884663e-05,
+      "loss": 2.7891,
+      "step": 10706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.545126353790614e-05,
+      "loss": 1.9173,
+      "step": 10708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5446635193927614e-05,
+      "loss": 0.3199,
+      "step": 10710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.544200684994909e-05,
+      "loss": 0.6595,
+      "step": 10712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5437378505970565e-05,
+      "loss": 0.3747,
+      "step": 10714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.543275016199204e-05,
+      "loss": 0.0018,
+      "step": 10716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5428121818013516e-05,
+      "loss": 1.6034,
+      "step": 10718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.542349347403499e-05,
+      "loss": 0.9929,
+      "step": 10720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5418865130056467e-05,
+      "loss": 2.8125,
+      "step": 10722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5414236786077946e-05,
+      "loss": 1.8049,
+      "step": 10724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.540960844209942e-05,
+      "loss": 1.3999,
+      "step": 10726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5404980098120896e-05,
+      "loss": 2.3082,
+      "step": 10728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5400351754142372e-05,
+      "loss": 4.2842,
+      "step": 10730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5395723410163847e-05,
+      "loss": 2.0197,
+      "step": 10732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5391095066185323e-05,
+      "loss": 5.4691,
+      "step": 10734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5386466722206798e-05,
+      "loss": 0.0421,
+      "step": 10736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5381838378228274e-05,
+      "loss": 4.4251,
+      "step": 10738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.537721003424975e-05,
+      "loss": 0.1543,
+      "step": 10740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5372581690271224e-05,
+      "loss": 1.2435,
+      "step": 10742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.53679533462927e-05,
+      "loss": 0.1654,
+      "step": 10744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5363325002314175e-05,
+      "loss": 2.1057,
+      "step": 10746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.535869665833565e-05,
+      "loss": 2.0205,
+      "step": 10748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5354068314357126e-05,
+      "loss": 0.0004,
+      "step": 10750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.53494399703786e-05,
+      "loss": 0.5455,
+      "step": 10752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5344811626400077e-05,
+      "loss": 0.6389,
+      "step": 10754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5340183282421552e-05,
+      "loss": 3.7369,
+      "step": 10756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5335554938443024e-05,
+      "loss": 0.0049,
+      "step": 10758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.53309265944645e-05,
+      "loss": 0.7895,
+      "step": 10760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5326298250485975e-05,
+      "loss": 2.4411,
+      "step": 10762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.532166990650745e-05,
+      "loss": 1.3168,
+      "step": 10764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5317041562528926e-05,
+      "loss": 0.2204,
+      "step": 10766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.53124132185504e-05,
+      "loss": 1.9499,
+      "step": 10768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5307784874571877e-05,
+      "loss": 7.521,
+      "step": 10770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5303156530593352e-05,
+      "loss": 0.0206,
+      "step": 10772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5298528186614828e-05,
+      "loss": 6.26,
+      "step": 10774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5293899842636303e-05,
+      "loss": 0.2263,
+      "step": 10776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.528927149865778e-05,
+      "loss": 1.5064,
+      "step": 10778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5284643154679254e-05,
+      "loss": 6.4635,
+      "step": 10780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.528001481070073e-05,
+      "loss": 0.8544,
+      "step": 10782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5275386466722205e-05,
+      "loss": 1.4406,
+      "step": 10784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.527075812274368e-05,
+      "loss": 1.1079,
+      "step": 10786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5266129778765156e-05,
+      "loss": 0.0105,
+      "step": 10788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5261501434786635e-05,
+      "loss": 0.3791,
+      "step": 10790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.525687309080811e-05,
+      "loss": 0.1399,
+      "step": 10792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5252244746829585e-05,
+      "loss": 5.3318,
+      "step": 10794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.524761640285106e-05,
+      "loss": 1.1062,
+      "step": 10796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5242988058872536e-05,
+      "loss": 3.1185,
+      "step": 10798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5238359714894012e-05,
+      "loss": 0.7466,
+      "step": 10800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5233731370915487e-05,
+      "loss": 3.2531,
+      "step": 10802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5229103026936963e-05,
+      "loss": 5.683,
+      "step": 10804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5224474682958438e-05,
+      "loss": 1.8788,
+      "step": 10806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5219846338979913e-05,
+      "loss": 4.1938,
+      "step": 10808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.521521799500139e-05,
+      "loss": 0.6317,
+      "step": 10810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5210589651022864e-05,
+      "loss": 0.6382,
+      "step": 10812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.520596130704434e-05,
+      "loss": 1.8167,
+      "step": 10814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5201332963065815e-05,
+      "loss": 3.7617,
+      "step": 10816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.519670461908729e-05,
+      "loss": 0.4565,
+      "step": 10818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5192076275108766e-05,
+      "loss": 8.7492,
+      "step": 10820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.518744793113024e-05,
+      "loss": 4.4655,
+      "step": 10822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5182819587151717e-05,
+      "loss": 2.2671,
+      "step": 10824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5178191243173192e-05,
+      "loss": 5.5665,
+      "step": 10826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5173562899194668e-05,
+      "loss": 1.1654,
+      "step": 10828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5168934555216147e-05,
+      "loss": 1.8174,
+      "step": 10830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5164306211237622e-05,
+      "loss": 2.59,
+      "step": 10832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5159677867259097e-05,
+      "loss": 1.6071,
+      "step": 10834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5155049523280573e-05,
+      "loss": 0.6106,
+      "step": 10836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5150421179302048e-05,
+      "loss": 4.189,
+      "step": 10838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5145792835323524e-05,
+      "loss": 2.3909,
+      "step": 10840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5141164491345e-05,
+      "loss": 0.5402,
+      "step": 10842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5136536147366474e-05,
+      "loss": 1.583,
+      "step": 10844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.513190780338795e-05,
+      "loss": 1.6838,
+      "step": 10846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5127279459409425e-05,
+      "loss": 0.6073,
+      "step": 10848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.51226511154309e-05,
+      "loss": 0.9323,
+      "step": 10850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.5118022771452376e-05,
+      "loss": 1.0693,
+      "step": 10852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.511339442747385e-05,
+      "loss": 3.8696,
+      "step": 10854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5108766083495327e-05,
+      "loss": 0.0855,
+      "step": 10856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5104137739516802e-05,
+      "loss": 1.5782,
+      "step": 10858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5099509395538278e-05,
+      "loss": 0.5446,
+      "step": 10860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5094881051559753e-05,
+      "loss": 1.9495,
+      "step": 10862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.509025270758123e-05,
+      "loss": 1.6402,
+      "step": 10864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5085624363602704e-05,
+      "loss": 1.0244,
+      "step": 10866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.508099601962418e-05,
+      "loss": 2.1611,
+      "step": 10868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.507636767564566e-05,
+      "loss": 0.9849,
+      "step": 10870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5071739331667134e-05,
+      "loss": 4.6912,
+      "step": 10872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.506711098768861e-05,
+      "loss": 0.0966,
+      "step": 10874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5062482643710085e-05,
+      "loss": 0.6236,
+      "step": 10876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.505785429973156e-05,
+      "loss": 0.0915,
+      "step": 10878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5053225955753036e-05,
+      "loss": 0.0418,
+      "step": 10880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.504859761177451e-05,
+      "loss": 4.8764,
+      "step": 10882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5043969267795986e-05,
+      "loss": 0.0274,
+      "step": 10884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5039340923817462e-05,
+      "loss": 0.9711,
+      "step": 10886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5034712579838937e-05,
+      "loss": 3.4981,
+      "step": 10888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5030084235860413e-05,
+      "loss": 4.7794,
+      "step": 10890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5025455891881888e-05,
+      "loss": 2.7099,
+      "step": 10892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5020827547903363e-05,
+      "loss": 8.029,
+      "step": 10894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.501619920392484e-05,
+      "loss": 0.4904,
+      "step": 10896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5011570859946314e-05,
+      "loss": 1.0337,
+      "step": 10898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.500694251596779e-05,
+      "loss": 0.5419,
+      "step": 10900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5002314171989265e-05,
+      "loss": 0.8179,
+      "step": 10902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.499768582801074e-05,
+      "loss": 4.9519,
+      "step": 10904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4993057484032216e-05,
+      "loss": 5.257,
+      "step": 10906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.498842914005369e-05,
+      "loss": 1.3346,
+      "step": 10908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4983800796075167e-05,
+      "loss": 1.2401,
+      "step": 10910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4979172452096642e-05,
+      "loss": 0.4535,
+      "step": 10912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4974544108118118e-05,
+      "loss": 4.1053,
+      "step": 10914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4969915764139593e-05,
+      "loss": 0.9084,
+      "step": 10916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.496528742016107e-05,
+      "loss": 3.7412,
+      "step": 10918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4960659076182544e-05,
+      "loss": 1.8607,
+      "step": 10920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.495603073220402e-05,
+      "loss": 1.4363,
+      "step": 10922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4951402388225495e-05,
+      "loss": 2.905,
+      "step": 10924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.494677404424697e-05,
+      "loss": 0.9201,
+      "step": 10926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4942145700268446e-05,
+      "loss": 0.1906,
+      "step": 10928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.493751735628992e-05,
+      "loss": 0.7125,
+      "step": 10930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4932889012311397e-05,
+      "loss": 3.6161,
+      "step": 10932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4928260668332872e-05,
+      "loss": 2.9353,
+      "step": 10934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4923632324354347e-05,
+      "loss": 3.4616,
+      "step": 10936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4919003980375823e-05,
+      "loss": 0.2123,
+      "step": 10938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4914375636397298e-05,
+      "loss": 0.8262,
+      "step": 10940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4909747292418774e-05,
+      "loss": 0.2252,
+      "step": 10942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.490511894844025e-05,
+      "loss": 0.0018,
+      "step": 10944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4900490604461725e-05,
+      "loss": 4.9001,
+      "step": 10946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.48958622604832e-05,
+      "loss": 7.4127,
+      "step": 10948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4891233916504675e-05,
+      "loss": 1.472,
+      "step": 10950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.488660557252615e-05,
+      "loss": 5.6679,
+      "step": 10952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4881977228547626e-05,
+      "loss": 3.4975,
+      "step": 10954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.48773488845691e-05,
+      "loss": 1.5239,
+      "step": 10956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4872720540590577e-05,
+      "loss": 1.6431,
+      "step": 10958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4868092196612052e-05,
+      "loss": 5.719,
+      "step": 10960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4863463852633528e-05,
+      "loss": 0.79,
+      "step": 10962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4858835508655003e-05,
+      "loss": 1.7244,
+      "step": 10964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.485420716467648e-05,
+      "loss": 3.345,
+      "step": 10966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4849578820697954e-05,
+      "loss": 3.5074,
+      "step": 10968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.484495047671943e-05,
+      "loss": 0.6708,
+      "step": 10970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4840322132740905e-05,
+      "loss": 0.4075,
+      "step": 10972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.483569378876238e-05,
+      "loss": 0.5238,
+      "step": 10974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4831065444783856e-05,
+      "loss": 0.636,
+      "step": 10976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4826437100805335e-05,
+      "loss": 4.4648,
+      "step": 10978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.482180875682681e-05,
+      "loss": 7.7705,
+      "step": 10980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4817180412848286e-05,
+      "loss": 1.9336,
+      "step": 10982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.481255206886976e-05,
+      "loss": 2.8995,
+      "step": 10984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4807923724891236e-05,
+      "loss": 0.4243,
+      "step": 10986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4803295380912712e-05,
+      "loss": 0.1339,
+      "step": 10988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4798667036934187e-05,
+      "loss": 0.4538,
+      "step": 10990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4794038692955663e-05,
+      "loss": 1.1609,
+      "step": 10992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4789410348977138e-05,
+      "loss": 1.7119,
+      "step": 10994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4784782004998614e-05,
+      "loss": 2.3233,
+      "step": 10996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.478015366102009e-05,
+      "loss": 1.4601,
+      "step": 10998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4775525317041564e-05,
+      "loss": 0.1126,
+      "step": 11000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.477089697306304e-05,
+      "loss": 1.3814,
+      "step": 11002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4766268629084515e-05,
+      "loss": 3.9005,
+      "step": 11004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.476164028510599e-05,
+      "loss": 1.4116,
+      "step": 11006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4757011941127466e-05,
+      "loss": 0.0004,
+      "step": 11008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.475238359714894e-05,
+      "loss": 1.5694,
+      "step": 11010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4747755253170417e-05,
+      "loss": 1.1881,
+      "step": 11012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4743126909191892e-05,
+      "loss": 1.2821,
+      "step": 11014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4738498565213368e-05,
+      "loss": 3.0448,
+      "step": 11016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4733870221234843e-05,
+      "loss": 3.5375,
+      "step": 11018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.472924187725632e-05,
+      "loss": 1.6291,
+      "step": 11020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4724613533277794e-05,
+      "loss": 3.6019,
+      "step": 11022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.471998518929927e-05,
+      "loss": 4.73,
+      "step": 11024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4715356845320745e-05,
+      "loss": 0.4036,
+      "step": 11026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.471072850134222e-05,
+      "loss": 0.3473,
+      "step": 11028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4706100157363696e-05,
+      "loss": 2.2996,
+      "step": 11030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.470147181338517e-05,
+      "loss": 2.5647,
+      "step": 11032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4696843469406647e-05,
+      "loss": 1.5384,
+      "step": 11034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4692215125428122e-05,
+      "loss": 0.062,
+      "step": 11036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4687586781449597e-05,
+      "loss": 0.0008,
+      "step": 11038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4682958437471073e-05,
+      "loss": 1.1076,
+      "step": 11040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4678330093492548e-05,
+      "loss": 5.0152,
+      "step": 11042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4673701749514024e-05,
+      "loss": 3.4418,
+      "step": 11044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.46690734055355e-05,
+      "loss": 2.503,
+      "step": 11046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4664445061556975e-05,
+      "loss": 1.4055,
+      "step": 11048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4659816717578453e-05,
+      "loss": 6.9918,
+      "step": 11050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.465518837359993e-05,
+      "loss": 1.0645,
+      "step": 11052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4650560029621404e-05,
+      "loss": 0.4999,
+      "step": 11054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.464593168564288e-05,
+      "loss": 0.0068,
+      "step": 11056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4641303341664355e-05,
+      "loss": 1.9184,
+      "step": 11058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.463667499768583e-05,
+      "loss": 5.7846,
+      "step": 11060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4632046653707306e-05,
+      "loss": 1.9106,
+      "step": 11062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.462741830972878e-05,
+      "loss": 0.9288,
+      "step": 11064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4622789965750257e-05,
+      "loss": 0.2789,
+      "step": 11066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4618161621771732e-05,
+      "loss": 1.1125,
+      "step": 11068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4613533277793208e-05,
+      "loss": 3.0751,
+      "step": 11070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4608904933814683e-05,
+      "loss": 0.3909,
+      "step": 11072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.460427658983616e-05,
+      "loss": 2.9488,
+      "step": 11074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4599648245857634e-05,
+      "loss": 2.9328,
+      "step": 11076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.459501990187911e-05,
+      "loss": 4.6936,
+      "step": 11078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4590391557900585e-05,
+      "loss": 1.6501,
+      "step": 11080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.458576321392206e-05,
+      "loss": 1.4187,
+      "step": 11082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4581134869943536e-05,
+      "loss": 2.6703,
+      "step": 11084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.457650652596501e-05,
+      "loss": 0.3244,
+      "step": 11086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4571878181986486e-05,
+      "loss": 4.3483,
+      "step": 11088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4567249838007962e-05,
+      "loss": 1.9387,
+      "step": 11090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4562621494029437e-05,
+      "loss": 2.4125,
+      "step": 11092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4557993150050913e-05,
+      "loss": 1.9836,
+      "step": 11094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4553364806072388e-05,
+      "loss": 1.4088,
+      "step": 11096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4548736462093864e-05,
+      "loss": 0.1005,
+      "step": 11098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.454410811811534e-05,
+      "loss": 3.8122,
+      "step": 11100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4539479774136814e-05,
+      "loss": 1.5739,
+      "step": 11102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.453485143015829e-05,
+      "loss": 4.0685,
+      "step": 11104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4530223086179765e-05,
+      "loss": 0.0224,
+      "step": 11106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.452559474220124e-05,
+      "loss": 4.1641,
+      "step": 11108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4520966398222716e-05,
+      "loss": 1.3253,
+      "step": 11110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.451633805424419e-05,
+      "loss": 6.5043,
+      "step": 11112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4511709710265667e-05,
+      "loss": 0.9125,
+      "step": 11114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4507081366287142e-05,
+      "loss": 3.5488,
+      "step": 11116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4502453022308618e-05,
+      "loss": 0.7689,
+      "step": 11118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4497824678330093e-05,
+      "loss": 1.1446,
+      "step": 11120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.449319633435157e-05,
+      "loss": 9.1204,
+      "step": 11122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4488567990373047e-05,
+      "loss": 1.1954,
+      "step": 11124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4483939646394523e-05,
+      "loss": 0.4774,
+      "step": 11126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4479311302416e-05,
+      "loss": 1.1023,
+      "step": 11128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4474682958437474e-05,
+      "loss": 2.4272,
+      "step": 11130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.447005461445895e-05,
+      "loss": 1.8325,
+      "step": 11132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4465426270480425e-05,
+      "loss": 3.4507,
+      "step": 11134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.44607979265019e-05,
+      "loss": 0.3498,
+      "step": 11136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4456169582523375e-05,
+      "loss": 1.2935,
+      "step": 11138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.445154123854485e-05,
+      "loss": 0.4234,
+      "step": 11140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4446912894566326e-05,
+      "loss": 3.0161,
+      "step": 11142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4442284550587802e-05,
+      "loss": 2.6918,
+      "step": 11144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4437656206609277e-05,
+      "loss": 0.7827,
+      "step": 11146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4433027862630753e-05,
+      "loss": 1.3444,
+      "step": 11148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4428399518652228e-05,
+      "loss": 1.5864,
+      "step": 11150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4423771174673703e-05,
+      "loss": 1.1552,
+      "step": 11152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.441914283069518e-05,
+      "loss": 3.0909,
+      "step": 11154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4414514486716654e-05,
+      "loss": 1.6787,
+      "step": 11156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.440988614273813e-05,
+      "loss": 1.4011,
+      "step": 11158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4405257798759605e-05,
+      "loss": 0.4605,
+      "step": 11160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.440062945478108e-05,
+      "loss": 1.6797,
+      "step": 11162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4396001110802556e-05,
+      "loss": 3.6558,
+      "step": 11164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.439137276682403e-05,
+      "loss": 0.7617,
+      "step": 11166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4386744422845507e-05,
+      "loss": 2.3279,
+      "step": 11168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4382116078866982e-05,
+      "loss": 1.2202,
+      "step": 11170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4377487734888458e-05,
+      "loss": 5.5606,
+      "step": 11172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4372859390909933e-05,
+      "loss": 1.1671,
+      "step": 11174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.436823104693141e-05,
+      "loss": 1.4214,
+      "step": 11176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4363602702952884e-05,
+      "loss": 0.0108,
+      "step": 11178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.435897435897436e-05,
+      "loss": 0.0096,
+      "step": 11180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4354346014995835e-05,
+      "loss": 9.1698,
+      "step": 11182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.434971767101731e-05,
+      "loss": 3.0189,
+      "step": 11184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4345089327038786e-05,
+      "loss": 2.0515,
+      "step": 11186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.434046098306026e-05,
+      "loss": 0.8165,
+      "step": 11188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4335832639081736e-05,
+      "loss": 2.0105,
+      "step": 11190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4331204295103212e-05,
+      "loss": 2.287,
+      "step": 11192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4326575951124687e-05,
+      "loss": 4.134,
+      "step": 11194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4321947607146163e-05,
+      "loss": 1.9106,
+      "step": 11196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.431731926316764e-05,
+      "loss": 3.4879,
+      "step": 11198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4312690919189117e-05,
+      "loss": 4.2655,
+      "step": 11200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4308062575210592e-05,
+      "loss": 1.7051,
+      "step": 11202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4303434231232068e-05,
+      "loss": 0.6493,
+      "step": 11204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4298805887253543e-05,
+      "loss": 0.2021,
+      "step": 11206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.429417754327502e-05,
+      "loss": 1.7346,
+      "step": 11208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4289549199296494e-05,
+      "loss": 0.7857,
+      "step": 11210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.428492085531797e-05,
+      "loss": 2.1978,
+      "step": 11212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4280292511339445e-05,
+      "loss": 1.2276,
+      "step": 11214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.427566416736092e-05,
+      "loss": 2.8579,
+      "step": 11216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4271035823382396e-05,
+      "loss": 4.1516,
+      "step": 11218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.426640747940387e-05,
+      "loss": 2.4513,
+      "step": 11220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4261779135425347e-05,
+      "loss": 0.0045,
+      "step": 11222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4257150791446822e-05,
+      "loss": 4.3487,
+      "step": 11224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4252522447468298e-05,
+      "loss": 1.3267,
+      "step": 11226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4247894103489773e-05,
+      "loss": 0.9638,
+      "step": 11228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.424326575951125e-05,
+      "loss": 1.8809,
+      "step": 11230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4238637415532724e-05,
+      "loss": 1.6923,
+      "step": 11232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.42340090715542e-05,
+      "loss": 0.0938,
+      "step": 11234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4229380727575675e-05,
+      "loss": 0.5582,
+      "step": 11236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.422475238359715e-05,
+      "loss": 0.6677,
+      "step": 11238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4220124039618625e-05,
+      "loss": 1.8938,
+      "step": 11240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.42154956956401e-05,
+      "loss": 0.2298,
+      "step": 11242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4210867351661576e-05,
+      "loss": 4.7688,
+      "step": 11244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4206239007683052e-05,
+      "loss": 2.1021,
+      "step": 11246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4201610663704527e-05,
+      "loss": 4.8994,
+      "step": 11248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4196982319726003e-05,
+      "loss": 0.1836,
+      "step": 11250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4192353975747478e-05,
+      "loss": 2.2905,
+      "step": 11252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4187725631768953e-05,
+      "loss": 2.5842,
+      "step": 11254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.418309728779043e-05,
+      "loss": 3.5241,
+      "step": 11256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4178468943811904e-05,
+      "loss": 0.538,
+      "step": 11258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.417384059983338e-05,
+      "loss": 0.7227,
+      "step": 11260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4169212255854855e-05,
+      "loss": 2.627,
+      "step": 11262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.416458391187633e-05,
+      "loss": 0.2266,
+      "step": 11264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4159955567897806e-05,
+      "loss": 2.3895,
+      "step": 11266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.415532722391928e-05,
+      "loss": 0.0019,
+      "step": 11268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4150698879940757e-05,
+      "loss": 1.3735,
+      "step": 11270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4146070535962236e-05,
+      "loss": 2.5134,
+      "step": 11272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.414144219198371e-05,
+      "loss": 3.7204,
+      "step": 11274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4136813848005187e-05,
+      "loss": 1.5583,
+      "step": 11276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4132185504026662e-05,
+      "loss": 4.004,
+      "step": 11278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4127557160048137e-05,
+      "loss": 0.0321,
+      "step": 11280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4122928816069613e-05,
+      "loss": 1.0213,
+      "step": 11282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4118300472091088e-05,
+      "loss": 0.0752,
+      "step": 11284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4113672128112564e-05,
+      "loss": 0.2098,
+      "step": 11286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.410904378413404e-05,
+      "loss": 0.029,
+      "step": 11288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4104415440155514e-05,
+      "loss": 2.3837,
+      "step": 11290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.409978709617699e-05,
+      "loss": 5.4758,
+      "step": 11292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4095158752198465e-05,
+      "loss": 0.0235,
+      "step": 11294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.409053040821994e-05,
+      "loss": 2.3618,
+      "step": 11296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4085902064241416e-05,
+      "loss": 2.8237,
+      "step": 11298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.408127372026289e-05,
+      "loss": 5.1632,
+      "step": 11300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4076645376284364e-05,
+      "loss": 3.2161,
+      "step": 11302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4072017032305842e-05,
+      "loss": 5.8994,
+      "step": 11304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4067388688327318e-05,
+      "loss": 1.0395,
+      "step": 11306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4062760344348793e-05,
+      "loss": 0.8664,
+      "step": 11308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.405813200037027e-05,
+      "loss": 3.0925,
+      "step": 11310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4053503656391744e-05,
+      "loss": 0.0007,
+      "step": 11312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.404887531241322e-05,
+      "loss": 2.9255,
+      "step": 11314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4044246968434695e-05,
+      "loss": 0.8451,
+      "step": 11316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.403961862445617e-05,
+      "loss": 0.9974,
+      "step": 11318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4034990280477646e-05,
+      "loss": 0.4491,
+      "step": 11320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.403036193649912e-05,
+      "loss": 4.5837,
+      "step": 11322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4025733592520597e-05,
+      "loss": 0.0209,
+      "step": 11324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4021105248542072e-05,
+      "loss": 0.8152,
+      "step": 11326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4016476904563548e-05,
+      "loss": 2.4669,
+      "step": 11328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4011848560585023e-05,
+      "loss": 1.0523,
+      "step": 11330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.40072202166065e-05,
+      "loss": 2.2443,
+      "step": 11332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.4002591872627974e-05,
+      "loss": 3.2072,
+      "step": 11334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.399796352864945e-05,
+      "loss": 0.8773,
+      "step": 11336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3993335184670925e-05,
+      "loss": 6.5746,
+      "step": 11338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.39887068406924e-05,
+      "loss": 2.7629,
+      "step": 11340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3984078496713876e-05,
+      "loss": 4.9417,
+      "step": 11342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3979450152735354e-05,
+      "loss": 1.7322,
+      "step": 11344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.397482180875683e-05,
+      "loss": 1.1291,
+      "step": 11346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3970193464778305e-05,
+      "loss": 0.7873,
+      "step": 11348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.396556512079978e-05,
+      "loss": 1.149,
+      "step": 11350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3960936776821256e-05,
+      "loss": 3.3263,
+      "step": 11352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.395630843284273e-05,
+      "loss": 3.1071,
+      "step": 11354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3951680088864207e-05,
+      "loss": 1.3147,
+      "step": 11356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3947051744885682e-05,
+      "loss": 2.9403,
+      "step": 11358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3942423400907158e-05,
+      "loss": 5.2262,
+      "step": 11360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3937795056928633e-05,
+      "loss": 2.6371,
+      "step": 11362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.393316671295011e-05,
+      "loss": 1.627,
+      "step": 11364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3928538368971584e-05,
+      "loss": 1.5642,
+      "step": 11366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.392391002499306e-05,
+      "loss": 2.2996,
+      "step": 11368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3919281681014535e-05,
+      "loss": 2.8258,
+      "step": 11370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.391465333703601e-05,
+      "loss": 1.2265,
+      "step": 11372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3910024993057486e-05,
+      "loss": 3.4283,
+      "step": 11374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3905396649078958e-05,
+      "loss": 2.0581,
+      "step": 11376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3900768305100437e-05,
+      "loss": 0.5135,
+      "step": 11378
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3896139961121912e-05,
+      "loss": 0.0293,
+      "step": 11380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3891511617143387e-05,
+      "loss": 0.8802,
+      "step": 11382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3886883273164863e-05,
+      "loss": 0.0109,
+      "step": 11384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3882254929186338e-05,
+      "loss": 2.0542,
+      "step": 11386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3877626585207814e-05,
+      "loss": 3.0506,
+      "step": 11388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.387299824122929e-05,
+      "loss": 3.9052,
+      "step": 11390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3868369897250765e-05,
+      "loss": 2.5502,
+      "step": 11392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.386374155327224e-05,
+      "loss": 1.7489,
+      "step": 11394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3859113209293715e-05,
+      "loss": 4.8033,
+      "step": 11396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.385448486531519e-05,
+      "loss": 1.0647,
+      "step": 11398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3849856521336666e-05,
+      "loss": 1.4736,
+      "step": 11400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.384522817735814e-05,
+      "loss": 3.6563,
+      "step": 11402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3840599833379617e-05,
+      "loss": 0.4296,
+      "step": 11404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3835971489401093e-05,
+      "loss": 4.3785,
+      "step": 11406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3831343145422568e-05,
+      "loss": 0.1951,
+      "step": 11408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3826714801444043e-05,
+      "loss": 2.6543,
+      "step": 11410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.382208645746552e-05,
+      "loss": 1.4794,
+      "step": 11412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3817458113486994e-05,
+      "loss": 4.8881,
+      "step": 11414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.381282976950847e-05,
+      "loss": 3.9938,
+      "step": 11416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.380820142552995e-05,
+      "loss": 1.8005,
+      "step": 11418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3803573081551424e-05,
+      "loss": 2.5097,
+      "step": 11420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.37989447375729e-05,
+      "loss": 1.7501,
+      "step": 11422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3794316393594375e-05,
+      "loss": 1.7017,
+      "step": 11424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.378968804961585e-05,
+      "loss": 2.2042,
+      "step": 11426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3785059705637326e-05,
+      "loss": 2.4984,
+      "step": 11428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.37804313616588e-05,
+      "loss": 2.1332,
+      "step": 11430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3775803017680276e-05,
+      "loss": 2.3465,
+      "step": 11432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3771174673701752e-05,
+      "loss": 0.4042,
+      "step": 11434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3766546329723227e-05,
+      "loss": 0.0968,
+      "step": 11436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3761917985744703e-05,
+      "loss": 1.6538,
+      "step": 11438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3757289641766178e-05,
+      "loss": 1.3352,
+      "step": 11440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3752661297787654e-05,
+      "loss": 1.3705,
+      "step": 11442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.374803295380913e-05,
+      "loss": 3.4098,
+      "step": 11444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3743404609830604e-05,
+      "loss": 1.3245,
+      "step": 11446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3738776265852076e-05,
+      "loss": 0.0375,
+      "step": 11448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3734147921873552e-05,
+      "loss": 0.3716,
+      "step": 11450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.372951957789503e-05,
+      "loss": 1.5376,
+      "step": 11452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3724891233916506e-05,
+      "loss": 0.5546,
+      "step": 11454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.372026288993798e-05,
+      "loss": 1.0139,
+      "step": 11456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3715634545959457e-05,
+      "loss": 0.0023,
+      "step": 11458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3711006201980932e-05,
+      "loss": 3.0647,
+      "step": 11460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3706377858002408e-05,
+      "loss": 3.8631,
+      "step": 11462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3701749514023883e-05,
+      "loss": 0.6844,
+      "step": 11464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.369712117004536e-05,
+      "loss": 0.3766,
+      "step": 11466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3692492826066834e-05,
+      "loss": 2.7217,
+      "step": 11468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.368786448208831e-05,
+      "loss": 0.6464,
+      "step": 11470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3683236138109785e-05,
+      "loss": 1.7295,
+      "step": 11472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.367860779413126e-05,
+      "loss": 1.1876,
+      "step": 11474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3673979450152736e-05,
+      "loss": 1.2564,
+      "step": 11476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.366935110617421e-05,
+      "loss": 1.8553,
+      "step": 11478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3664722762195687e-05,
+      "loss": 0.5782,
+      "step": 11480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3660094418217162e-05,
+      "loss": 2.9281,
+      "step": 11482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3655466074238637e-05,
+      "loss": 3.0873,
+      "step": 11484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3650837730260113e-05,
+      "loss": 2.3988,
+      "step": 11486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3646209386281588e-05,
+      "loss": 0.8782,
+      "step": 11488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3641581042303064e-05,
+      "loss": 0.538,
+      "step": 11490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3636952698324543e-05,
+      "loss": 3.4558,
+      "step": 11492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3632324354346018e-05,
+      "loss": 0.8473,
+      "step": 11494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3627696010367493e-05,
+      "loss": 4.7022,
+      "step": 11496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.362306766638897e-05,
+      "loss": 2.7692,
+      "step": 11498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3618439322410444e-05,
+      "loss": 0.2777,
+      "step": 11500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.361381097843192e-05,
+      "loss": 2.711,
+      "step": 11502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3609182634453395e-05,
+      "loss": 0.3319,
+      "step": 11504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.360455429047487e-05,
+      "loss": 2.4296,
+      "step": 11506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3599925946496346e-05,
+      "loss": 0.0285,
+      "step": 11508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.359529760251782e-05,
+      "loss": 0.8064,
+      "step": 11510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3590669258539297e-05,
+      "loss": 0.7026,
+      "step": 11512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3586040914560772e-05,
+      "loss": 2.4599,
+      "step": 11514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3581412570582248e-05,
+      "loss": 3.1368,
+      "step": 11516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3576784226603723e-05,
+      "loss": 0.6655,
+      "step": 11518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3572155882625195e-05,
+      "loss": 1.0324,
+      "step": 11520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.356752753864667e-05,
+      "loss": 2.6682,
+      "step": 11522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.356289919466815e-05,
+      "loss": 4.3664,
+      "step": 11524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3558270850689625e-05,
+      "loss": 3.0065,
+      "step": 11526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.35536425067111e-05,
+      "loss": 2.6501,
+      "step": 11528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3549014162732576e-05,
+      "loss": 3.4268,
+      "step": 11530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.354438581875405e-05,
+      "loss": 1.523,
+      "step": 11532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3539757474775526e-05,
+      "loss": 0.0009,
+      "step": 11534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3535129130797002e-05,
+      "loss": 0.3034,
+      "step": 11536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3530500786818477e-05,
+      "loss": 3.6636,
+      "step": 11538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3525872442839953e-05,
+      "loss": 0.1919,
+      "step": 11540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3521244098861428e-05,
+      "loss": 4.7357,
+      "step": 11542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3516615754882904e-05,
+      "loss": 4.7137,
+      "step": 11544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.351198741090438e-05,
+      "loss": 1.358,
+      "step": 11546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3507359066925854e-05,
+      "loss": 0.7011,
+      "step": 11548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.350273072294733e-05,
+      "loss": 1.4574,
+      "step": 11550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3498102378968805e-05,
+      "loss": 0.2736,
+      "step": 11552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.349347403499028e-05,
+      "loss": 0.1344,
+      "step": 11554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3488845691011756e-05,
+      "loss": 1.2024,
+      "step": 11556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.348421734703323e-05,
+      "loss": 0.0404,
+      "step": 11558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3479589003054707e-05,
+      "loss": 0.33,
+      "step": 11560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3474960659076182e-05,
+      "loss": 0.7041,
+      "step": 11562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3470332315097658e-05,
+      "loss": 10.2638,
+      "step": 11564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3465703971119137e-05,
+      "loss": 7.7568,
+      "step": 11566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3461075627140612e-05,
+      "loss": 0.0003,
+      "step": 11568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3456447283162087e-05,
+      "loss": 2.2152,
+      "step": 11570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3451818939183563e-05,
+      "loss": 4.1176,
+      "step": 11572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.344719059520504e-05,
+      "loss": 2.023,
+      "step": 11574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3442562251226514e-05,
+      "loss": 5.7705,
+      "step": 11576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.343793390724799e-05,
+      "loss": 3.1709,
+      "step": 11578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3433305563269465e-05,
+      "loss": 0.0009,
+      "step": 11580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.342867721929094e-05,
+      "loss": 2.233,
+      "step": 11582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3424048875312415e-05,
+      "loss": 4.4059,
+      "step": 11584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.341942053133389e-05,
+      "loss": 1.0962,
+      "step": 11586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3414792187355366e-05,
+      "loss": 1.2199,
+      "step": 11588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3410163843376842e-05,
+      "loss": 0.0006,
+      "step": 11590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3405535499398314e-05,
+      "loss": 0.0079,
+      "step": 11592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.340090715541979e-05,
+      "loss": 0.9957,
+      "step": 11594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3396278811441265e-05,
+      "loss": 0.0022,
+      "step": 11596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3391650467462743e-05,
+      "loss": 0.7252,
+      "step": 11598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.338702212348422e-05,
+      "loss": 1.7499,
+      "step": 11600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3382393779505694e-05,
+      "loss": 1.8626,
+      "step": 11602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.337776543552717e-05,
+      "loss": 3.66,
+      "step": 11604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3373137091548645e-05,
+      "loss": 4.4469,
+      "step": 11606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.336850874757012e-05,
+      "loss": 0.0276,
+      "step": 11608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3363880403591596e-05,
+      "loss": 1.7478,
+      "step": 11610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.335925205961307e-05,
+      "loss": 2.1135,
+      "step": 11612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3354623715634547e-05,
+      "loss": 0.6441,
+      "step": 11614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3349995371656022e-05,
+      "loss": 2.9487,
+      "step": 11616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3345367027677498e-05,
+      "loss": 0.7412,
+      "step": 11618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3340738683698973e-05,
+      "loss": 5.2675,
+      "step": 11620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.333611033972045e-05,
+      "loss": 1.0765,
+      "step": 11622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3331481995741924e-05,
+      "loss": 1.6913,
+      "step": 11624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.33268536517634e-05,
+      "loss": 6.9639,
+      "step": 11626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3322225307784875e-05,
+      "loss": 1.9966,
+      "step": 11628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.331759696380635e-05,
+      "loss": 5.2761,
+      "step": 11630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3312968619827826e-05,
+      "loss": 0.9856,
+      "step": 11632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.33083402758493e-05,
+      "loss": 0.5208,
+      "step": 11634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3303711931870776e-05,
+      "loss": 0.0003,
+      "step": 11636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3299083587892255e-05,
+      "loss": 3.4098,
+      "step": 11638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.329445524391373e-05,
+      "loss": 2.8005,
+      "step": 11640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3289826899935206e-05,
+      "loss": 2.3016,
+      "step": 11642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.328519855595668e-05,
+      "loss": 1.4691,
+      "step": 11644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3280570211978157e-05,
+      "loss": 2.1602,
+      "step": 11646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3275941867999632e-05,
+      "loss": 3.6909,
+      "step": 11648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3271313524021108e-05,
+      "loss": 4.2358,
+      "step": 11650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3266685180042583e-05,
+      "loss": 1.8802,
+      "step": 11652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.326205683606406e-05,
+      "loss": 0.6052,
+      "step": 11654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3257428492085534e-05,
+      "loss": 2.2932,
+      "step": 11656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.325280014810701e-05,
+      "loss": 1.8191,
+      "step": 11658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3248171804128485e-05,
+      "loss": 0.9598,
+      "step": 11660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.324354346014996e-05,
+      "loss": 0.0234,
+      "step": 11662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3238915116171436e-05,
+      "loss": 0.7169,
+      "step": 11664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3234286772192908e-05,
+      "loss": 1.2792,
+      "step": 11666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3229658428214383e-05,
+      "loss": 0.7113,
+      "step": 11668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.322503008423586e-05,
+      "loss": 0.0011,
+      "step": 11670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3220401740257338e-05,
+      "loss": 3.6669,
+      "step": 11672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3215773396278813e-05,
+      "loss": 2.6983,
+      "step": 11674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.321114505230029e-05,
+      "loss": 2.6359,
+      "step": 11676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3206516708321764e-05,
+      "loss": 3.3102,
+      "step": 11678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.320188836434324e-05,
+      "loss": 0.0005,
+      "step": 11680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3197260020364715e-05,
+      "loss": 3.6713,
+      "step": 11682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.319263167638619e-05,
+      "loss": 10.8348,
+      "step": 11684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3188003332407665e-05,
+      "loss": 0.0024,
+      "step": 11686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.318337498842914e-05,
+      "loss": 5.8768,
+      "step": 11688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3178746644450616e-05,
+      "loss": 1.3058,
+      "step": 11690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3174118300472092e-05,
+      "loss": 3.7066,
+      "step": 11692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3169489956493567e-05,
+      "loss": 2.9691,
+      "step": 11694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3164861612515043e-05,
+      "loss": 1.0647,
+      "step": 11696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3160233268536518e-05,
+      "loss": 3.6425,
+      "step": 11698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3155604924557993e-05,
+      "loss": 1.4321,
+      "step": 11700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.315097658057947e-05,
+      "loss": 2.86,
+      "step": 11702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3146348236600944e-05,
+      "loss": 0.9618,
+      "step": 11704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.314171989262242e-05,
+      "loss": 2.982,
+      "step": 11706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3137091548643895e-05,
+      "loss": 2.3021,
+      "step": 11708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.313246320466537e-05,
+      "loss": 2.0069,
+      "step": 11710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.312783486068685e-05,
+      "loss": 0.0012,
+      "step": 11712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3123206516708325e-05,
+      "loss": 0.4581,
+      "step": 11714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.31185781727298e-05,
+      "loss": 1.3937,
+      "step": 11716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.3113949828751276e-05,
+      "loss": 7.0825,
+      "step": 11718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.310932148477275e-05,
+      "loss": 2.146,
+      "step": 11720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3104693140794227e-05,
+      "loss": 1.3761,
+      "step": 11722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3100064796815702e-05,
+      "loss": 4.7641,
+      "step": 11724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3095436452837177e-05,
+      "loss": 1.0774,
+      "step": 11726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3090808108858653e-05,
+      "loss": 0.2275,
+      "step": 11728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3086179764880128e-05,
+      "loss": 0.3565,
+      "step": 11730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3081551420901604e-05,
+      "loss": 0.9936,
+      "step": 11732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.6058,
+      "step": 11734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3072294732944554e-05,
+      "loss": 2.5883,
+      "step": 11736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3067666388966027e-05,
+      "loss": 3.8657,
+      "step": 11738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3063038044987502e-05,
+      "loss": 2.1753,
+      "step": 11740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3058409701008977e-05,
+      "loss": 0.0013,
+      "step": 11742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3053781357030453e-05,
+      "loss": 4.6145,
+      "step": 11744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.304915301305193e-05,
+      "loss": 1.5403,
+      "step": 11746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3044524669073407e-05,
+      "loss": 1.677,
+      "step": 11748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3039896325094882e-05,
+      "loss": 2.9177,
+      "step": 11750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3035267981116358e-05,
+      "loss": 2.2892,
+      "step": 11752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3030639637137833e-05,
+      "loss": 5.2272,
+      "step": 11754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.302601129315931e-05,
+      "loss": 1.6585,
+      "step": 11756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3021382949180784e-05,
+      "loss": 1.8311,
+      "step": 11758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.301675460520226e-05,
+      "loss": 2.511,
+      "step": 11760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3012126261223735e-05,
+      "loss": 0.6595,
+      "step": 11762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.300749791724521e-05,
+      "loss": 0.0007,
+      "step": 11764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.3002869573266686e-05,
+      "loss": 2.5219,
+      "step": 11766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.299824122928816e-05,
+      "loss": 1.1799,
+      "step": 11768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2993612885309637e-05,
+      "loss": 3.7213,
+      "step": 11770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2988984541331112e-05,
+      "loss": 2.2108,
+      "step": 11772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2984356197352588e-05,
+      "loss": 3.0596,
+      "step": 11774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2979727853374063e-05,
+      "loss": 5.0035,
+      "step": 11776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.297509950939554e-05,
+      "loss": 0.8988,
+      "step": 11778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2970471165417014e-05,
+      "loss": 3.5068,
+      "step": 11780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.296584282143849e-05,
+      "loss": 5.0437,
+      "step": 11782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2961214477459965e-05,
+      "loss": 1.7216,
+      "step": 11784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2956586133481443e-05,
+      "loss": 2.3211,
+      "step": 11786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.295195778950292e-05,
+      "loss": 2.3235,
+      "step": 11788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2947329445524394e-05,
+      "loss": 0.1274,
+      "step": 11790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.294270110154587e-05,
+      "loss": 0.001,
+      "step": 11792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2938072757567345e-05,
+      "loss": 2.5048,
+      "step": 11794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.293344441358882e-05,
+      "loss": 0.006,
+      "step": 11796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2928816069610296e-05,
+      "loss": 1.2315,
+      "step": 11798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.292418772563177e-05,
+      "loss": 0.4612,
+      "step": 11800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2919559381653247e-05,
+      "loss": 0.0338,
+      "step": 11802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2914931037674722e-05,
+      "loss": 0.4349,
+      "step": 11804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2910302693696198e-05,
+      "loss": 0.5837,
+      "step": 11806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2905674349717673e-05,
+      "loss": 1.2775,
+      "step": 11808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2901046005739145e-05,
+      "loss": 1.5945,
+      "step": 11810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.289641766176062e-05,
+      "loss": 1.0095,
+      "step": 11812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2891789317782096e-05,
+      "loss": 0.8258,
+      "step": 11814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.288716097380357e-05,
+      "loss": 1.6127,
+      "step": 11816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.288253262982505e-05,
+      "loss": 3.479,
+      "step": 11818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2877904285846526e-05,
+      "loss": 9.351,
+      "step": 11820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2873275941868e-05,
+      "loss": 0.0002,
+      "step": 11822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2868647597889477e-05,
+      "loss": 4.8827,
+      "step": 11824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2864019253910952e-05,
+      "loss": 2.687,
+      "step": 11826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2859390909932427e-05,
+      "loss": 2.8048,
+      "step": 11828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2854762565953903e-05,
+      "loss": 1.8187,
+      "step": 11830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2850134221975378e-05,
+      "loss": 1.8712,
+      "step": 11832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2845505877996854e-05,
+      "loss": 5.4094,
+      "step": 11834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.284087753401833e-05,
+      "loss": 3.6381,
+      "step": 11836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2836249190039805e-05,
+      "loss": 3.7207,
+      "step": 11838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.283162084606128e-05,
+      "loss": 3.2465,
+      "step": 11840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2826992502082755e-05,
+      "loss": 5.7429,
+      "step": 11842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.282236415810423e-05,
+      "loss": 0.4259,
+      "step": 11844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2817735814125706e-05,
+      "loss": 3.4085,
+      "step": 11846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.281310747014718e-05,
+      "loss": 2.4615,
+      "step": 11848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2808479126168657e-05,
+      "loss": 0.7117,
+      "step": 11850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2803850782190133e-05,
+      "loss": 0.9904,
+      "step": 11852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2799222438211608e-05,
+      "loss": 0.401,
+      "step": 11854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2794594094233083e-05,
+      "loss": 1.3235,
+      "step": 11856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2789965750254562e-05,
+      "loss": 0.2817,
+      "step": 11858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2785337406276038e-05,
+      "loss": 1.772,
+      "step": 11860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2780709062297513e-05,
+      "loss": 1.1804,
+      "step": 11862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.277608071831899e-05,
+      "loss": 4.126,
+      "step": 11864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2771452374340464e-05,
+      "loss": 2.1236,
+      "step": 11866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.276682403036194e-05,
+      "loss": 2.2264,
+      "step": 11868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2762195686383415e-05,
+      "loss": 2.2909,
+      "step": 11870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.275756734240489e-05,
+      "loss": 1.2049,
+      "step": 11872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2752938998426366e-05,
+      "loss": 3.8166,
+      "step": 11874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.274831065444784e-05,
+      "loss": 0.7682,
+      "step": 11876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2743682310469316e-05,
+      "loss": 2.1898,
+      "step": 11878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2739053966490792e-05,
+      "loss": 0.0312,
+      "step": 11880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2734425622512264e-05,
+      "loss": 2.8355,
+      "step": 11882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.272979727853374e-05,
+      "loss": 2.6993,
+      "step": 11884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2725168934555215e-05,
+      "loss": 2.3613,
+      "step": 11886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.272054059057669e-05,
+      "loss": 2.2117,
+      "step": 11888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2715912246598166e-05,
+      "loss": 2.7101,
+      "step": 11890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2711283902619644e-05,
+      "loss": 1.9092,
+      "step": 11892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.270665555864112e-05,
+      "loss": 1.6353,
+      "step": 11894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2702027214662595e-05,
+      "loss": 1.932,
+      "step": 11896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.269739887068407e-05,
+      "loss": 2.0946,
+      "step": 11898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2692770526705546e-05,
+      "loss": 1.0947,
+      "step": 11900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.268814218272702e-05,
+      "loss": 2.2807,
+      "step": 11902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2683513838748497e-05,
+      "loss": 0.9438,
+      "step": 11904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2678885494769972e-05,
+      "loss": 0.0048,
+      "step": 11906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2674257150791448e-05,
+      "loss": 4.6263,
+      "step": 11908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2669628806812923e-05,
+      "loss": 0.6084,
+      "step": 11910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.26650004628344e-05,
+      "loss": 1.6278,
+      "step": 11912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2660372118855874e-05,
+      "loss": 1.5517,
+      "step": 11914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.265574377487735e-05,
+      "loss": 0.6334,
+      "step": 11916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2651115430898825e-05,
+      "loss": 2.9594,
+      "step": 11918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.26464870869203e-05,
+      "loss": 2.9019,
+      "step": 11920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2641858742941776e-05,
+      "loss": 3.7601,
+      "step": 11922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.263723039896325e-05,
+      "loss": 0.1491,
+      "step": 11924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2632602054984727e-05,
+      "loss": 0.8837,
+      "step": 11926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2627973711006202e-05,
+      "loss": 2.8138,
+      "step": 11928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2623345367027677e-05,
+      "loss": 3.7212,
+      "step": 11930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2618717023049156e-05,
+      "loss": 1.4218,
+      "step": 11932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.261408867907063e-05,
+      "loss": 1.7363,
+      "step": 11934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2609460335092107e-05,
+      "loss": 0.9116,
+      "step": 11936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2604831991113583e-05,
+      "loss": 0.0264,
+      "step": 11938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2600203647135058e-05,
+      "loss": 6.5916,
+      "step": 11940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2595575303156533e-05,
+      "loss": 4.521,
+      "step": 11942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.259094695917801e-05,
+      "loss": 0.9365,
+      "step": 11944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2586318615199484e-05,
+      "loss": 3.0691,
+      "step": 11946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.258169027122096e-05,
+      "loss": 0.3633,
+      "step": 11948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2577061927242435e-05,
+      "loss": 1.173,
+      "step": 11950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.257243358326391e-05,
+      "loss": 0.7139,
+      "step": 11952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2567805239285386e-05,
+      "loss": 0.0773,
+      "step": 11954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2563176895306858e-05,
+      "loss": 6.4459,
+      "step": 11956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2558548551328333e-05,
+      "loss": 0.0249,
+      "step": 11958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.255392020734981e-05,
+      "loss": 0.2342,
+      "step": 11960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2549291863371284e-05,
+      "loss": 0.5842,
+      "step": 11962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.254466351939276e-05,
+      "loss": 1.543,
+      "step": 11964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.254003517541424e-05,
+      "loss": 0.115,
+      "step": 11966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2535406831435714e-05,
+      "loss": 2.0399,
+      "step": 11968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.253077848745719e-05,
+      "loss": 0.1222,
+      "step": 11970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2526150143478665e-05,
+      "loss": 0.0018,
+      "step": 11972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.252152179950014e-05,
+      "loss": 0.1606,
+      "step": 11974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2516893455521616e-05,
+      "loss": 0.0932,
+      "step": 11976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.251226511154309e-05,
+      "loss": 2.6749,
+      "step": 11978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2507636767564566e-05,
+      "loss": 0.8746,
+      "step": 11980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2503008423586042e-05,
+      "loss": 1.6348,
+      "step": 11982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2498380079607517e-05,
+      "loss": 4.1786,
+      "step": 11984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2493751735628993e-05,
+      "loss": 5.1557,
+      "step": 11986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2489123391650468e-05,
+      "loss": 0.6608,
+      "step": 11988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2484495047671944e-05,
+      "loss": 4.0326,
+      "step": 11990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.247986670369342e-05,
+      "loss": 5.0737,
+      "step": 11992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2475238359714894e-05,
+      "loss": 0.0045,
+      "step": 11994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.247061001573637e-05,
+      "loss": 0.1323,
+      "step": 11996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2465981671757845e-05,
+      "loss": 5.4044,
+      "step": 11998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.246135332777932e-05,
+      "loss": 0.8646,
+      "step": 12000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2456724983800796e-05,
+      "loss": 3.5455,
+      "step": 12002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.245209663982227e-05,
+      "loss": 2.9853,
+      "step": 12004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.244746829584375e-05,
+      "loss": 3.3614,
+      "step": 12006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2442839951865226e-05,
+      "loss": 4.2385,
+      "step": 12008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.24382116078867e-05,
+      "loss": 0.2548,
+      "step": 12010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2433583263908177e-05,
+      "loss": 3.9689,
+      "step": 12012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2428954919929652e-05,
+      "loss": 1.7411,
+      "step": 12014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2424326575951127e-05,
+      "loss": 4.1931,
+      "step": 12016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2419698231972603e-05,
+      "loss": 2.3021,
+      "step": 12018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.241506988799408e-05,
+      "loss": 2.5676,
+      "step": 12020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2410441544015554e-05,
+      "loss": 2.5231,
+      "step": 12022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.240581320003703e-05,
+      "loss": 0.963,
+      "step": 12024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2401184856058505e-05,
+      "loss": 2.9893,
+      "step": 12026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2396556512079977e-05,
+      "loss": 3.1346,
+      "step": 12028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2391928168101452e-05,
+      "loss": 0.734,
+      "step": 12030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2387299824122927e-05,
+      "loss": 0.3818,
+      "step": 12032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2382671480144403e-05,
+      "loss": 0.7746,
+      "step": 12034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.237804313616588e-05,
+      "loss": 1.9022,
+      "step": 12036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2373414792187357e-05,
+      "loss": 0.9363,
+      "step": 12038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2368786448208833e-05,
+      "loss": 6.3397,
+      "step": 12040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2364158104230308e-05,
+      "loss": 0.6541,
+      "step": 12042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2359529760251783e-05,
+      "loss": 0.0231,
+      "step": 12044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.235490141627326e-05,
+      "loss": 1.9023,
+      "step": 12046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2350273072294734e-05,
+      "loss": 0.6167,
+      "step": 12048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.234564472831621e-05,
+      "loss": 3.7969,
+      "step": 12050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2341016384337685e-05,
+      "loss": 1.2143,
+      "step": 12052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.233638804035916e-05,
+      "loss": 1.7819,
+      "step": 12054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2331759696380636e-05,
+      "loss": 1.2891,
+      "step": 12056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.232713135240211e-05,
+      "loss": 0.0008,
+      "step": 12058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2322503008423587e-05,
+      "loss": 0.9438,
+      "step": 12060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2317874664445062e-05,
+      "loss": 0.3625,
+      "step": 12062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2313246320466538e-05,
+      "loss": 1.334,
+      "step": 12064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2308617976488013e-05,
+      "loss": 1.6971,
+      "step": 12066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.230398963250949e-05,
+      "loss": 6.2369,
+      "step": 12068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2299361288530964e-05,
+      "loss": 4.2267,
+      "step": 12070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.229473294455244e-05,
+      "loss": 4.8855,
+      "step": 12072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2290104600573915e-05,
+      "loss": 1.0188,
+      "step": 12074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.228547625659539e-05,
+      "loss": 0.0299,
+      "step": 12076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2280847912616866e-05,
+      "loss": 0.0009,
+      "step": 12078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2276219568638344e-05,
+      "loss": 2.502,
+      "step": 12080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.227159122465982e-05,
+      "loss": 1.8609,
+      "step": 12082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2266962880681295e-05,
+      "loss": 0.9015,
+      "step": 12084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.226233453670277e-05,
+      "loss": 2.3432,
+      "step": 12086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2257706192724246e-05,
+      "loss": 2.0292,
+      "step": 12088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.225307784874572e-05,
+      "loss": 0.0017,
+      "step": 12090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2248449504767197e-05,
+      "loss": 0.6692,
+      "step": 12092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2243821160788672e-05,
+      "loss": 0.0007,
+      "step": 12094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2239192816810148e-05,
+      "loss": 4.3866,
+      "step": 12096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2234564472831623e-05,
+      "loss": 0.4791,
+      "step": 12098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2229936128853095e-05,
+      "loss": 3.2548,
+      "step": 12100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.222530778487457e-05,
+      "loss": 3.4736,
+      "step": 12102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2220679440896046e-05,
+      "loss": 0.0005,
+      "step": 12104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.221605109691752e-05,
+      "loss": 1.9351,
+      "step": 12106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2211422752938997e-05,
+      "loss": 1.5958,
+      "step": 12108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2206794408960472e-05,
+      "loss": 2.1569,
+      "step": 12110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.220216606498195e-05,
+      "loss": 2.6058,
+      "step": 12112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2197537721003427e-05,
+      "loss": 3.3552,
+      "step": 12114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2192909377024902e-05,
+      "loss": 2.2583,
+      "step": 12116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2188281033046378e-05,
+      "loss": 2.1984,
+      "step": 12118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2183652689067853e-05,
+      "loss": 0.004,
+      "step": 12120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.217902434508933e-05,
+      "loss": 0.9704,
+      "step": 12122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2174396001110804e-05,
+      "loss": 1.2457,
+      "step": 12124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.216976765713228e-05,
+      "loss": 0.5382,
+      "step": 12126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2165139313153755e-05,
+      "loss": 5.6551,
+      "step": 12128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.216051096917523e-05,
+      "loss": 0.4806,
+      "step": 12130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2155882625196706e-05,
+      "loss": 0.0005,
+      "step": 12132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.215125428121818e-05,
+      "loss": 2.9863,
+      "step": 12134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2146625937239656e-05,
+      "loss": 1.9234,
+      "step": 12136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2141997593261132e-05,
+      "loss": 5.055,
+      "step": 12138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2137369249282607e-05,
+      "loss": 1.0453,
+      "step": 12140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2132740905304083e-05,
+      "loss": 1.0144,
+      "step": 12142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2128112561325558e-05,
+      "loss": 0.1294,
+      "step": 12144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2123484217347033e-05,
+      "loss": 2.2321,
+      "step": 12146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.211885587336851e-05,
+      "loss": 1.4939,
+      "step": 12148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2114227529389984e-05,
+      "loss": 8.8338,
+      "step": 12150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2109599185411463e-05,
+      "loss": 2.9383,
+      "step": 12152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.210497084143294e-05,
+      "loss": 1.5335,
+      "step": 12154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2100342497454414e-05,
+      "loss": 0.0331,
+      "step": 12156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.209571415347589e-05,
+      "loss": 0.9013,
+      "step": 12158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2091085809497365e-05,
+      "loss": 3.6362,
+      "step": 12160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.208645746551884e-05,
+      "loss": 1.8841,
+      "step": 12162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2081829121540316e-05,
+      "loss": 3.9153,
+      "step": 12164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.207720077756179e-05,
+      "loss": 3.5298,
+      "step": 12166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2072572433583267e-05,
+      "loss": 0.2455,
+      "step": 12168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2067944089604742e-05,
+      "loss": 5.2964,
+      "step": 12170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2063315745626214e-05,
+      "loss": 3.1358,
+      "step": 12172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.205868740164769e-05,
+      "loss": 1.8038,
+      "step": 12174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2054059057669165e-05,
+      "loss": 2.3105,
+      "step": 12176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.204943071369064e-05,
+      "loss": 1.283,
+      "step": 12178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2044802369712116e-05,
+      "loss": 0.8064,
+      "step": 12180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.204017402573359e-05,
+      "loss": 2.1813,
+      "step": 12182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2035545681755067e-05,
+      "loss": 4.4802,
+      "step": 12184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2030917337776545e-05,
+      "loss": 0.1084,
+      "step": 12186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.202628899379802e-05,
+      "loss": 0.9921,
+      "step": 12188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2021660649819496e-05,
+      "loss": 1.6508,
+      "step": 12190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.201703230584097e-05,
+      "loss": 1.4724,
+      "step": 12192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2012403961862447e-05,
+      "loss": 2.9722,
+      "step": 12194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2007775617883922e-05,
+      "loss": 1.4826,
+      "step": 12196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.2003147273905398e-05,
+      "loss": 0.2834,
+      "step": 12198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1998518929926873e-05,
+      "loss": 0.1618,
+      "step": 12200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.199389058594835e-05,
+      "loss": 5.1766,
+      "step": 12202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1989262241969824e-05,
+      "loss": 2.6722,
+      "step": 12204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.19846338979913e-05,
+      "loss": 0.0054,
+      "step": 12206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1980005554012775e-05,
+      "loss": 2.4434,
+      "step": 12208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.197537721003425e-05,
+      "loss": 2.4976,
+      "step": 12210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1970748866055726e-05,
+      "loss": 0.9715,
+      "step": 12212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.19661205220772e-05,
+      "loss": 3.3726,
+      "step": 12214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1961492178098677e-05,
+      "loss": 2.3836,
+      "step": 12216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1956863834120152e-05,
+      "loss": 3.7789,
+      "step": 12218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1952235490141628e-05,
+      "loss": 4.7102,
+      "step": 12220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1947607146163103e-05,
+      "loss": 0.8015,
+      "step": 12222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.194297880218458e-05,
+      "loss": 0.3311,
+      "step": 12224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1938350458206057e-05,
+      "loss": 2.3435,
+      "step": 12226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1933722114227533e-05,
+      "loss": 1.0832,
+      "step": 12228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1929093770249008e-05,
+      "loss": 3.1573,
+      "step": 12230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1924465426270484e-05,
+      "loss": 1.0696,
+      "step": 12232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.191983708229196e-05,
+      "loss": 0.7533,
+      "step": 12234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1915208738313434e-05,
+      "loss": 1.7905,
+      "step": 12236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.191058039433491e-05,
+      "loss": 1.5091,
+      "step": 12238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1905952050356385e-05,
+      "loss": 0.914,
+      "step": 12240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.190132370637786e-05,
+      "loss": 2.3226,
+      "step": 12242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1896695362399336e-05,
+      "loss": 0.0121,
+      "step": 12244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1892067018420808e-05,
+      "loss": 0.7989,
+      "step": 12246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1887438674442284e-05,
+      "loss": 1.2917,
+      "step": 12248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.188281033046376e-05,
+      "loss": 2.9295,
+      "step": 12250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1878181986485234e-05,
+      "loss": 3.944,
+      "step": 12252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.187355364250671e-05,
+      "loss": 3.2112,
+      "step": 12254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1868925298528185e-05,
+      "loss": 0.0018,
+      "step": 12256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.186429695454966e-05,
+      "loss": 5.8588,
+      "step": 12258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.185966861057114e-05,
+      "loss": 3.2808,
+      "step": 12260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1855040266592615e-05,
+      "loss": 1.3014,
+      "step": 12262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.185041192261409e-05,
+      "loss": 3.1334,
+      "step": 12264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1845783578635566e-05,
+      "loss": 1.7578,
+      "step": 12266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.184115523465704e-05,
+      "loss": 0.0008,
+      "step": 12268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1836526890678517e-05,
+      "loss": 1.9474,
+      "step": 12270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1831898546699992e-05,
+      "loss": 0.0259,
+      "step": 12272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1827270202721467e-05,
+      "loss": 0.1273,
+      "step": 12274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1822641858742943e-05,
+      "loss": 3.8764,
+      "step": 12276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1818013514764418e-05,
+      "loss": 1.2258,
+      "step": 12278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1813385170785894e-05,
+      "loss": 0.4149,
+      "step": 12280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.180875682680737e-05,
+      "loss": 1.7527,
+      "step": 12282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1804128482828845e-05,
+      "loss": 0.3528,
+      "step": 12284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.179950013885032e-05,
+      "loss": 1.575,
+      "step": 12286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1794871794871795e-05,
+      "loss": 0.2444,
+      "step": 12288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.179024345089327e-05,
+      "loss": 3.2597,
+      "step": 12290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1785615106914746e-05,
+      "loss": 0.5994,
+      "step": 12292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.178098676293622e-05,
+      "loss": 4.9125,
+      "step": 12294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1776358418957697e-05,
+      "loss": 0.0007,
+      "step": 12296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1771730074979173e-05,
+      "loss": 0.0036,
+      "step": 12298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.176710173100065e-05,
+      "loss": 0.0005,
+      "step": 12300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1762473387022127e-05,
+      "loss": 5.7649,
+      "step": 12302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1757845043043602e-05,
+      "loss": 3.4341,
+      "step": 12304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1753216699065078e-05,
+      "loss": 2.5572,
+      "step": 12306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1748588355086553e-05,
+      "loss": 2.3305,
+      "step": 12308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.174396001110803e-05,
+      "loss": 0.9504,
+      "step": 12310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1739331667129504e-05,
+      "loss": 6.5185,
+      "step": 12312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.173470332315098e-05,
+      "loss": 1.2865,
+      "step": 12314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1730074979172455e-05,
+      "loss": 1.8144,
+      "step": 12316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1725446635193927e-05,
+      "loss": 1.6653,
+      "step": 12318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1720818291215402e-05,
+      "loss": 1.0492,
+      "step": 12320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1716189947236878e-05,
+      "loss": 0.0015,
+      "step": 12322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1711561603258353e-05,
+      "loss": 1.3316,
+      "step": 12324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.170693325927983e-05,
+      "loss": 0.2044,
+      "step": 12326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1702304915301304e-05,
+      "loss": 0.0038,
+      "step": 12328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.169767657132278e-05,
+      "loss": 0.5461,
+      "step": 12330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1693048227344258e-05,
+      "loss": 0.7605,
+      "step": 12332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1688419883365734e-05,
+      "loss": 0.192,
+      "step": 12334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.168379153938721e-05,
+      "loss": 0.6901,
+      "step": 12336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1679163195408684e-05,
+      "loss": 0.0063,
+      "step": 12338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.167453485143016e-05,
+      "loss": 1.4803,
+      "step": 12340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1669906507451635e-05,
+      "loss": 0.6461,
+      "step": 12342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.166527816347311e-05,
+      "loss": 0.0063,
+      "step": 12344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1660649819494586e-05,
+      "loss": 10.8623,
+      "step": 12346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.165602147551606e-05,
+      "loss": 1.2709,
+      "step": 12348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1651393131537537e-05,
+      "loss": 0.1504,
+      "step": 12350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1646764787559012e-05,
+      "loss": 0.0179,
+      "step": 12352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1642136443580488e-05,
+      "loss": 1.2711,
+      "step": 12354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1637508099601963e-05,
+      "loss": 0.7502,
+      "step": 12356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.163287975562344e-05,
+      "loss": 2.5909,
+      "step": 12358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1628251411644914e-05,
+      "loss": 0.8392,
+      "step": 12360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.162362306766639e-05,
+      "loss": 0.9376,
+      "step": 12362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1618994723687865e-05,
+      "loss": 2.5356,
+      "step": 12364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.161436637970934e-05,
+      "loss": 0.044,
+      "step": 12366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1609738035730816e-05,
+      "loss": 0.7521,
+      "step": 12368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.160510969175229e-05,
+      "loss": 0.3094,
+      "step": 12370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1600481347773767e-05,
+      "loss": 8.6228,
+      "step": 12372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1595853003795245e-05,
+      "loss": 3.5834,
+      "step": 12374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.159122465981672e-05,
+      "loss": 2.2051,
+      "step": 12376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1586596315838196e-05,
+      "loss": 3.6073,
+      "step": 12378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1581967971859672e-05,
+      "loss": 0.0027,
+      "step": 12380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1577339627881147e-05,
+      "loss": 1.0955,
+      "step": 12382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1572711283902623e-05,
+      "loss": 2.8374,
+      "step": 12384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1568082939924098e-05,
+      "loss": 2.2274,
+      "step": 12386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1563454595945573e-05,
+      "loss": 0.0005,
+      "step": 12388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1558826251967045e-05,
+      "loss": 1.9176,
+      "step": 12390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.155419790798852e-05,
+      "loss": 2.9668,
+      "step": 12392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1549569564009996e-05,
+      "loss": 4.8629,
+      "step": 12394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1544941220031472e-05,
+      "loss": 5.0929,
+      "step": 12396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1540312876052947e-05,
+      "loss": 1.4477,
+      "step": 12398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1535684532074423e-05,
+      "loss": 1.8518,
+      "step": 12400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1531056188095898e-05,
+      "loss": 0.9887,
+      "step": 12402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1526427844117373e-05,
+      "loss": 0.3072,
+      "step": 12404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1521799500138852e-05,
+      "loss": 3.3984,
+      "step": 12406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1517171156160328e-05,
+      "loss": 1.8133,
+      "step": 12408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1512542812181803e-05,
+      "loss": 1.4735,
+      "step": 12410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.150791446820328e-05,
+      "loss": 1.2587,
+      "step": 12412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1503286124224754e-05,
+      "loss": 2.3244,
+      "step": 12414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.149865778024623e-05,
+      "loss": 0.1937,
+      "step": 12416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1494029436267705e-05,
+      "loss": 3.2552,
+      "step": 12418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.148940109228918e-05,
+      "loss": 3.808,
+      "step": 12420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1484772748310656e-05,
+      "loss": 0.9016,
+      "step": 12422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.148014440433213e-05,
+      "loss": 4.1214,
+      "step": 12424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1475516060353606e-05,
+      "loss": 0.0313,
+      "step": 12426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1470887716375082e-05,
+      "loss": 2.4133,
+      "step": 12428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1466259372396557e-05,
+      "loss": 1.5587,
+      "step": 12430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1461631028418033e-05,
+      "loss": 0.0006,
+      "step": 12432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1457002684439508e-05,
+      "loss": 0.0472,
+      "step": 12434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1452374340460984e-05,
+      "loss": 2.6021,
+      "step": 12436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.144774599648246e-05,
+      "loss": 1.9521,
+      "step": 12438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1443117652503934e-05,
+      "loss": 3.805,
+      "step": 12440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.143848930852541e-05,
+      "loss": 0.8187,
+      "step": 12442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1433860964546885e-05,
+      "loss": 0.7968,
+      "step": 12444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1429232620568364e-05,
+      "loss": 2.4827,
+      "step": 12446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142460427658984e-05,
+      "loss": 1.9531,
+      "step": 12448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1419975932611315e-05,
+      "loss": 2.2906,
+      "step": 12450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141534758863279e-05,
+      "loss": 0.0272,
+      "step": 12452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1410719244654266e-05,
+      "loss": 0.7308,
+      "step": 12454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140609090067574e-05,
+      "loss": 6.2819,
+      "step": 12456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1401462556697217e-05,
+      "loss": 4.7064,
+      "step": 12458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1396834212718692e-05,
+      "loss": 3.9976,
+      "step": 12460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1392205868740164e-05,
+      "loss": 0.4805,
+      "step": 12462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138757752476164e-05,
+      "loss": 1.328,
+      "step": 12464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1382949180783115e-05,
+      "loss": 0.0034,
+      "step": 12466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137832083680459e-05,
+      "loss": 0.0015,
+      "step": 12468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1373692492826066e-05,
+      "loss": 5.5032,
+      "step": 12470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136906414884754e-05,
+      "loss": 0.2279,
+      "step": 12472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1364435804869017e-05,
+      "loss": 3.3994,
+      "step": 12474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1359807460890492e-05,
+      "loss": 4.1042,
+      "step": 12476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1355179116911968e-05,
+      "loss": 0.4884,
+      "step": 12478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1350550772933446e-05,
+      "loss": 2.0963,
+      "step": 12480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1345922428954922e-05,
+      "loss": 3.9096,
+      "step": 12482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1341294084976397e-05,
+      "loss": 0.9575,
+      "step": 12484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1336665740997873e-05,
+      "loss": 1.9021,
+      "step": 12486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1332037397019348e-05,
+      "loss": 0.0673,
+      "step": 12488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1327409053040823e-05,
+      "loss": 0.0168,
+      "step": 12490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13227807090623e-05,
+      "loss": 0.5293,
+      "step": 12492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1318152365083774e-05,
+      "loss": 0.5195,
+      "step": 12494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131352402110525e-05,
+      "loss": 5.101,
+      "step": 12496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1308895677126725e-05,
+      "loss": 0.7056,
+      "step": 12498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13042673331482e-05,
+      "loss": 1.975,
+      "step": 12500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1299638989169676e-05,
+      "loss": 1.7308,
+      "step": 12502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.129501064519115e-05,
+      "loss": 0.001,
+      "step": 12504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1290382301212627e-05,
+      "loss": 0.5795,
+      "step": 12506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1285753957234102e-05,
+      "loss": 6.7808,
+      "step": 12508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1281125613255578e-05,
+      "loss": 0.737,
+      "step": 12510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1276497269277053e-05,
+      "loss": 2.0307,
+      "step": 12512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.127186892529853e-05,
+      "loss": 1.8205,
+      "step": 12514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1267240581320004e-05,
+      "loss": 7.3931,
+      "step": 12516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.126261223734148e-05,
+      "loss": 5.2621,
+      "step": 12518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1257983893362958e-05,
+      "loss": 4.605,
+      "step": 12520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1253355549384434e-05,
+      "loss": 0.0411,
+      "step": 12522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.124872720540591e-05,
+      "loss": 3.6925,
+      "step": 12524
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1244098861427384e-05,
+      "loss": 0.0049,
+      "step": 12526
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.123947051744886e-05,
+      "loss": 1.5443,
+      "step": 12528
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1234842173470335e-05,
+      "loss": 0.4193,
+      "step": 12530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.123021382949181e-05,
+      "loss": 0.619,
+      "step": 12532
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1225585485513286e-05,
+      "loss": 0.002,
+      "step": 12534
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1220957141534758e-05,
+      "loss": 0.0042,
+      "step": 12536
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1216328797556234e-05,
+      "loss": 2.2486,
+      "step": 12538
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.121170045357771e-05,
+      "loss": 0.2441,
+      "step": 12540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1207072109599184e-05,
+      "loss": 0.0034,
+      "step": 12542
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.120244376562066e-05,
+      "loss": 0.5065,
+      "step": 12544
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1197815421642135e-05,
+      "loss": 1.5904,
+      "step": 12546
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.119318707766361e-05,
+      "loss": 0.0006,
+      "step": 12548
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1188558733685086e-05,
+      "loss": 0.0007,
+      "step": 12550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.118393038970656e-05,
+      "loss": 3.5764,
+      "step": 12552
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.117930204572804e-05,
+      "loss": 0.0056,
+      "step": 12554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1174673701749516e-05,
+      "loss": 0.0141,
+      "step": 12556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.117004535777099e-05,
+      "loss": 0.7065,
+      "step": 12558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1165417013792467e-05,
+      "loss": 0.4386,
+      "step": 12560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1160788669813942e-05,
+      "loss": 0.0003,
+      "step": 12562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1156160325835418e-05,
+      "loss": 0.0003,
+      "step": 12564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1151531981856893e-05,
+      "loss": 3.5506,
+      "step": 12566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.114690363787837e-05,
+      "loss": 0.0942,
+      "step": 12568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1142275293899844e-05,
+      "loss": 2.3554,
+      "step": 12570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.113764694992132e-05,
+      "loss": 0.009,
+      "step": 12572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1133018605942795e-05,
+      "loss": 0.9075,
+      "step": 12574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.112839026196427e-05,
+      "loss": 0.0004,
+      "step": 12576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1123761917985746e-05,
+      "loss": 3.1724,
+      "step": 12578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.111913357400722e-05,
+      "loss": 2.4001,
+      "step": 12580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1114505230028696e-05,
+      "loss": 3.1454,
+      "step": 12582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1109876886050172e-05,
+      "loss": 5.189,
+      "step": 12584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1105248542071647e-05,
+      "loss": 1.067,
+      "step": 12586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1100620198093123e-05,
+      "loss": 2.3112,
+      "step": 12588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1095991854114598e-05,
+      "loss": 4.5257,
+      "step": 12590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1091363510136073e-05,
+      "loss": 2.5536,
+      "step": 12592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1086735166157552e-05,
+      "loss": 3.8379,
+      "step": 12594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1082106822179028e-05,
+      "loss": 0.3602,
+      "step": 12596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1077478478200503e-05,
+      "loss": 0.0007,
+      "step": 12598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.107285013422198e-05,
+      "loss": 0.8321,
+      "step": 12600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1068221790243454e-05,
+      "loss": 0.0004,
+      "step": 12602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.106359344626493e-05,
+      "loss": 3.036,
+      "step": 12604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1058965102286405e-05,
+      "loss": 0.3945,
+      "step": 12606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1054336758307877e-05,
+      "loss": 0.7987,
+      "step": 12608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1049708414329352e-05,
+      "loss": 3.4054,
+      "step": 12610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1045080070350828e-05,
+      "loss": 2.5751,
+      "step": 12612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1040451726372303e-05,
+      "loss": 3.7509,
+      "step": 12614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.103582338239378e-05,
+      "loss": 1.5793,
+      "step": 12616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1031195038415254e-05,
+      "loss": 2.4525,
+      "step": 12618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.102656669443673e-05,
+      "loss": 4.0913,
+      "step": 12620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1021938350458205e-05,
+      "loss": 2.4787,
+      "step": 12622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.101731000647968e-05,
+      "loss": 0.036,
+      "step": 12624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.101268166250116e-05,
+      "loss": 0.0006,
+      "step": 12626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1008053318522635e-05,
+      "loss": 1.1443,
+      "step": 12628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.100342497454411e-05,
+      "loss": 1.4905,
+      "step": 12630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0998796630565585e-05,
+      "loss": 0.1112,
+      "step": 12632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.099416828658706e-05,
+      "loss": 5.1063,
+      "step": 12634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0989539942608536e-05,
+      "loss": 0.6458,
+      "step": 12636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.098491159863001e-05,
+      "loss": 2.1376,
+      "step": 12638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0980283254651487e-05,
+      "loss": 2.3098,
+      "step": 12640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0975654910672962e-05,
+      "loss": 2.2796,
+      "step": 12642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0971026566694438e-05,
+      "loss": 1.2828,
+      "step": 12644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0966398222715913e-05,
+      "loss": 1.5862,
+      "step": 12646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.096176987873739e-05,
+      "loss": 0.5897,
+      "step": 12648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0957141534758864e-05,
+      "loss": 9.0643,
+      "step": 12650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.095251319078034e-05,
+      "loss": 2.4861,
+      "step": 12652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0947884846801815e-05,
+      "loss": 0.0003,
+      "step": 12654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.094325650282329e-05,
+      "loss": 7.8567,
+      "step": 12656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0938628158844766e-05,
+      "loss": 0.9355,
+      "step": 12658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.093399981486624e-05,
+      "loss": 0.7322,
+      "step": 12660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0929371470887717e-05,
+      "loss": 1.5692,
+      "step": 12662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0924743126909192e-05,
+      "loss": 0.4464,
+      "step": 12664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.092011478293067e-05,
+      "loss": 1.1605,
+      "step": 12666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0915486438952146e-05,
+      "loss": 0.7542,
+      "step": 12668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0910858094973622e-05,
+      "loss": 4.1189,
+      "step": 12670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0906229750995097e-05,
+      "loss": 1.3662,
+      "step": 12672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0901601407016573e-05,
+      "loss": 5.5894,
+      "step": 12674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0896973063038048e-05,
+      "loss": 0.2732,
+      "step": 12676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0892344719059524e-05,
+      "loss": 3.548,
+      "step": 12678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0887716375080996e-05,
+      "loss": 1.5096,
+      "step": 12680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.088308803110247e-05,
+      "loss": 0.2838,
+      "step": 12682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0878459687123946e-05,
+      "loss": 1.7244,
+      "step": 12684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0873831343145422e-05,
+      "loss": 2.6493,
+      "step": 12686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0869202999166897e-05,
+      "loss": 2.255,
+      "step": 12688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0864574655188373e-05,
+      "loss": 1.7124,
+      "step": 12690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0859946311209848e-05,
+      "loss": 0.8541,
+      "step": 12692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0855317967231324e-05,
+      "loss": 4.0551,
+      "step": 12694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.08506896232528e-05,
+      "loss": 2.8005,
+      "step": 12696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0846061279274274e-05,
+      "loss": 2.2532,
+      "step": 12698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0841432935295753e-05,
+      "loss": 0.0196,
+      "step": 12700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.083680459131723e-05,
+      "loss": 0.1673,
+      "step": 12702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0832176247338704e-05,
+      "loss": 3.9807,
+      "step": 12704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.082754790336018e-05,
+      "loss": 0.0847,
+      "step": 12706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0822919559381655e-05,
+      "loss": 0.4479,
+      "step": 12708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.081829121540313e-05,
+      "loss": 2.2053,
+      "step": 12710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0813662871424606e-05,
+      "loss": 2.7647,
+      "step": 12712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.080903452744608e-05,
+      "loss": 0.8371,
+      "step": 12714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0804406183467557e-05,
+      "loss": 2.5045,
+      "step": 12716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0799777839489032e-05,
+      "loss": 1.2528,
+      "step": 12718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0795149495510507e-05,
+      "loss": 2.1941,
+      "step": 12720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0790521151531983e-05,
+      "loss": 0.214,
+      "step": 12722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0785892807553458e-05,
+      "loss": 1.1561,
+      "step": 12724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0781264463574934e-05,
+      "loss": 0.8329,
+      "step": 12726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.077663611959641e-05,
+      "loss": 0.337,
+      "step": 12728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0772007775617885e-05,
+      "loss": 1.6052,
+      "step": 12730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.076737943163936e-05,
+      "loss": 1.4785,
+      "step": 12732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0762751087660835e-05,
+      "loss": 0.1332,
+      "step": 12734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.075812274368231e-05,
+      "loss": 3.4472,
+      "step": 12736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0753494399703786e-05,
+      "loss": 1.1413,
+      "step": 12738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0748866055725265e-05,
+      "loss": 1.5087,
+      "step": 12740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.074423771174674e-05,
+      "loss": 1.7845,
+      "step": 12742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0739609367768216e-05,
+      "loss": 5.8833,
+      "step": 12744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.073498102378969e-05,
+      "loss": 0.2313,
+      "step": 12746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0730352679811167e-05,
+      "loss": 0.0004,
+      "step": 12748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0725724335832642e-05,
+      "loss": 1.6512,
+      "step": 12750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0721095991854114e-05,
+      "loss": 2.8994,
+      "step": 12752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.071646764787559e-05,
+      "loss": 1.3225,
+      "step": 12754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0711839303897065e-05,
+      "loss": 3.3233,
+      "step": 12756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.070721095991854e-05,
+      "loss": 0.049,
+      "step": 12758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0702582615940016e-05,
+      "loss": 4.8666,
+      "step": 12760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.069795427196149e-05,
+      "loss": 0.0006,
+      "step": 12762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0693325927982967e-05,
+      "loss": 0.0006,
+      "step": 12764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0688697584004442e-05,
+      "loss": 1.4813,
+      "step": 12766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0684069240025918e-05,
+      "loss": 6.5309,
+      "step": 12768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0679440896047393e-05,
+      "loss": 4.0542,
+      "step": 12770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.067481255206887e-05,
+      "loss": 1.0355,
+      "step": 12772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0670184208090347e-05,
+      "loss": 4.674,
+      "step": 12774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0665555864111823e-05,
+      "loss": 4.5592,
+      "step": 12776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0660927520133298e-05,
+      "loss": 0.3153,
+      "step": 12778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0656299176154774e-05,
+      "loss": 0.1512,
+      "step": 12780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.065167083217625e-05,
+      "loss": 4.1385,
+      "step": 12782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0647042488197724e-05,
+      "loss": 3.406,
+      "step": 12784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.06424141442192e-05,
+      "loss": 1.9741,
+      "step": 12786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0637785800240675e-05,
+      "loss": 7.9059,
+      "step": 12788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.063315745626215e-05,
+      "loss": 0.0782,
+      "step": 12790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0628529112283626e-05,
+      "loss": 0.9353,
+      "step": 12792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.06239007683051e-05,
+      "loss": 0.0064,
+      "step": 12794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0619272424326577e-05,
+      "loss": 1.4788,
+      "step": 12796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0614644080348052e-05,
+      "loss": 0.0015,
+      "step": 12798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0610015736369528e-05,
+      "loss": 2.7884,
+      "step": 12800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0605387392391003e-05,
+      "loss": 0.8954,
+      "step": 12802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.060075904841248e-05,
+      "loss": 0.9421,
+      "step": 12804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0596130704433954e-05,
+      "loss": 0.0039,
+      "step": 12806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.059150236045543e-05,
+      "loss": 2.3414,
+      "step": 12808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0586874016476905e-05,
+      "loss": 0.6735,
+      "step": 12810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.058224567249838e-05,
+      "loss": 3.2192,
+      "step": 12812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.057761732851986e-05,
+      "loss": 1.1327,
+      "step": 12814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0572988984541335e-05,
+      "loss": 4.704,
+      "step": 12816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.056836064056281e-05,
+      "loss": 1.83,
+      "step": 12818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0563732296584285e-05,
+      "loss": 4.1857,
+      "step": 12820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.055910395260576e-05,
+      "loss": 1.3008,
+      "step": 12822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0554475608627236e-05,
+      "loss": 1.7161,
+      "step": 12824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.054984726464871e-05,
+      "loss": 2.7708,
+      "step": 12826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0545218920670184e-05,
+      "loss": 2.8979,
+      "step": 12828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.054059057669166e-05,
+      "loss": 1.6352,
+      "step": 12830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0535962232713135e-05,
+      "loss": 4.6781,
+      "step": 12832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.053133388873461e-05,
+      "loss": 0.8877,
+      "step": 12834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0526705544756085e-05,
+      "loss": 0.4359,
+      "step": 12836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.052207720077756e-05,
+      "loss": 3.532,
+      "step": 12838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0517448856799036e-05,
+      "loss": 2.4899,
+      "step": 12840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 0.0106,
+      "step": 12842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0508192168841987e-05,
+      "loss": 0.6557,
+      "step": 12844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0503563824863466e-05,
+      "loss": 0.0064,
+      "step": 12846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.049893548088494e-05,
+      "loss": 0.8528,
+      "step": 12848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0494307136906417e-05,
+      "loss": 0.8456,
+      "step": 12850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0489678792927892e-05,
+      "loss": 3.2293,
+      "step": 12852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0485050448949368e-05,
+      "loss": 2.3659,
+      "step": 12854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0480422104970843e-05,
+      "loss": 1.6216,
+      "step": 12856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.047579376099232e-05,
+      "loss": 0.3492,
+      "step": 12858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0471165417013794e-05,
+      "loss": 6.1425,
+      "step": 12860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.046653707303527e-05,
+      "loss": 4.7499,
+      "step": 12862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0461908729056745e-05,
+      "loss": 0.0539,
+      "step": 12864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.045728038507822e-05,
+      "loss": 0.977,
+      "step": 12866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0452652041099696e-05,
+      "loss": 0.0017,
+      "step": 12868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.044802369712117e-05,
+      "loss": 1.4585,
+      "step": 12870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0443395353142646e-05,
+      "loss": 2.122,
+      "step": 12872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0438767009164122e-05,
+      "loss": 2.2445,
+      "step": 12874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0434138665185597e-05,
+      "loss": 0.2004,
+      "step": 12876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0429510321207073e-05,
+      "loss": 0.433,
+      "step": 12878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0424881977228548e-05,
+      "loss": 1.3428,
+      "step": 12880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0420253633250024e-05,
+      "loss": 1.0899,
+      "step": 12882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.04156252892715e-05,
+      "loss": 0.0003,
+      "step": 12884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0410996945292974e-05,
+      "loss": 8.1004,
+      "step": 12886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0406368601314453e-05,
+      "loss": 2.4207,
+      "step": 12888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.040174025733593e-05,
+      "loss": 1.0712,
+      "step": 12890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0397111913357404e-05,
+      "loss": 1.9182,
+      "step": 12892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.039248356937888e-05,
+      "loss": 1.3091,
+      "step": 12894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0387855225400355e-05,
+      "loss": 3.1954,
+      "step": 12896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0383226881421827e-05,
+      "loss": 1.1417,
+      "step": 12898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0378598537443302e-05,
+      "loss": 8.073,
+      "step": 12900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0373970193464778e-05,
+      "loss": 1.351,
+      "step": 12902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0369341849486253e-05,
+      "loss": 0.9674,
+      "step": 12904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.036471350550773e-05,
+      "loss": 3.9606,
+      "step": 12906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0360085161529204e-05,
+      "loss": 4.4583,
+      "step": 12908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.035545681755068e-05,
+      "loss": 0.8235,
+      "step": 12910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0350828473572155e-05,
+      "loss": 0.885,
+      "step": 12912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.034620012959363e-05,
+      "loss": 2.0684,
+      "step": 12914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0341571785615106e-05,
+      "loss": 0.8671,
+      "step": 12916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.033694344163658e-05,
+      "loss": 2.2294,
+      "step": 12918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.033231509765806e-05,
+      "loss": 2.7857,
+      "step": 12920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0327686753679535e-05,
+      "loss": 2.0503,
+      "step": 12922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.032305840970101e-05,
+      "loss": 1.6263,
+      "step": 12924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0318430065722486e-05,
+      "loss": 1.1333,
+      "step": 12926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0313801721743962e-05,
+      "loss": 0.4096,
+      "step": 12928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0309173377765437e-05,
+      "loss": 2.0053,
+      "step": 12930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0304545033786913e-05,
+      "loss": 2.9911,
+      "step": 12932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0299916689808388e-05,
+      "loss": 1.2863,
+      "step": 12934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0295288345829863e-05,
+      "loss": 2.0375,
+      "step": 12936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.029066000185134e-05,
+      "loss": 1.662,
+      "step": 12938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0286031657872814e-05,
+      "loss": 0.5018,
+      "step": 12940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.028140331389429e-05,
+      "loss": 2.9333,
+      "step": 12942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0276774969915765e-05,
+      "loss": 0.1588,
+      "step": 12944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.027214662593724e-05,
+      "loss": 0.1355,
+      "step": 12946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0267518281958716e-05,
+      "loss": 3.4959,
+      "step": 12948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.026288993798019e-05,
+      "loss": 0.0731,
+      "step": 12950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0258261594001667e-05,
+      "loss": 0.0023,
+      "step": 12952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0253633250023142e-05,
+      "loss": 5.8339,
+      "step": 12954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0249004906044618e-05,
+      "loss": 3.2694,
+      "step": 12956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0244376562066093e-05,
+      "loss": 2.7101,
+      "step": 12958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0239748218087572e-05,
+      "loss": 2.5793,
+      "step": 12960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0235119874109047e-05,
+      "loss": 0.3884,
+      "step": 12962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0230491530130523e-05,
+      "loss": 0.4251,
+      "step": 12964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0225863186151998e-05,
+      "loss": 5.3666,
+      "step": 12966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0221234842173474e-05,
+      "loss": 0.0722,
+      "step": 12968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0216606498194946e-05,
+      "loss": 2.2163,
+      "step": 12970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.021197815421642e-05,
+      "loss": 1.84,
+      "step": 12972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0207349810237897e-05,
+      "loss": 6.5032,
+      "step": 12974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0202721466259372e-05,
+      "loss": 0.5512,
+      "step": 12976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0198093122280847e-05,
+      "loss": 4.4299,
+      "step": 12978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0193464778302323e-05,
+      "loss": 6.1811,
+      "step": 12980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0188836434323798e-05,
+      "loss": 0.5635,
+      "step": 12982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0184208090345274e-05,
+      "loss": 2.2466,
+      "step": 12984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.017957974636675e-05,
+      "loss": 2.2934,
+      "step": 12986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0174951402388224e-05,
+      "loss": 2.1685,
+      "step": 12988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.01703230584097e-05,
+      "loss": 0.7945,
+      "step": 12990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0165694714431175e-05,
+      "loss": 1.0952,
+      "step": 12992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0161066370452654e-05,
+      "loss": 3.53,
+      "step": 12994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.015643802647413e-05,
+      "loss": 4.7015,
+      "step": 12996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0151809682495605e-05,
+      "loss": 3.2434,
+      "step": 12998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.014718133851708e-05,
+      "loss": 3.6138,
+      "step": 13000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0142552994538556e-05,
+      "loss": 4.4853,
+      "step": 13002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.013792465056003e-05,
+      "loss": 3.1557,
+      "step": 13004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0133296306581507e-05,
+      "loss": 2.6206,
+      "step": 13006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0128667962602982e-05,
+      "loss": 1.1429,
+      "step": 13008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0124039618624458e-05,
+      "loss": 3.8085,
+      "step": 13010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0119411274645933e-05,
+      "loss": 0.3173,
+      "step": 13012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.011478293066741e-05,
+      "loss": 0.9372,
+      "step": 13014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0110154586688884e-05,
+      "loss": 0.9879,
+      "step": 13016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.010552624271036e-05,
+      "loss": 2.7954,
+      "step": 13018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0100897898731835e-05,
+      "loss": 1.0263,
+      "step": 13020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.009626955475331e-05,
+      "loss": 3.327,
+      "step": 13022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0091641210774786e-05,
+      "loss": 4.23,
+      "step": 13024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.008701286679626e-05,
+      "loss": 4.3586,
+      "step": 13026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0082384522817736e-05,
+      "loss": 1.4732,
+      "step": 13028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0077756178839212e-05,
+      "loss": 0.9702,
+      "step": 13030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0073127834860687e-05,
+      "loss": 0.1685,
+      "step": 13032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0068499490882166e-05,
+      "loss": 0.2289,
+      "step": 13034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.006387114690364e-05,
+      "loss": 1.0722,
+      "step": 13036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0059242802925117e-05,
+      "loss": 0.558,
+      "step": 13038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0054614458946592e-05,
+      "loss": 0.0638,
+      "step": 13040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0049986114968064e-05,
+      "loss": 2.7206,
+      "step": 13042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.004535777098954e-05,
+      "loss": 0.8041,
+      "step": 13044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0040729427011015e-05,
+      "loss": 0.0004,
+      "step": 13046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.003610108303249e-05,
+      "loss": 2.5999,
+      "step": 13048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0031472739053966e-05,
+      "loss": 2.5038,
+      "step": 13050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.002684439507544e-05,
+      "loss": 3.0792,
+      "step": 13052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0022216051096917e-05,
+      "loss": 1.8828,
+      "step": 13054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0017587707118392e-05,
+      "loss": 4.3957,
+      "step": 13056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0012959363139868e-05,
+      "loss": 4.3117,
+      "step": 13058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0008331019161343e-05,
+      "loss": 0.0086,
+      "step": 13060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.000370267518282e-05,
+      "loss": 0.0156,
+      "step": 13062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999074331204294e-05,
+      "loss": 2.0186,
+      "step": 13064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999444598722577e-05,
+      "loss": 5.6924,
+      "step": 13066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9989817643247248e-05,
+      "loss": 0.3991,
+      "step": 13068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9985189299268724e-05,
+      "loss": 5.2702,
+      "step": 13070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99805609552902e-05,
+      "loss": 0.446,
+      "step": 13072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975932611311675e-05,
+      "loss": 0.1372,
+      "step": 13074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997130426733315e-05,
+      "loss": 0.0803,
+      "step": 13076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9966675923354625e-05,
+      "loss": 1.4658,
+      "step": 13078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99620475793761e-05,
+      "loss": 2.6716,
+      "step": 13080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9957419235397576e-05,
+      "loss": 1.9509,
+      "step": 13082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.995279089141905e-05,
+      "loss": 2.1457,
+      "step": 13084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9948162547440527e-05,
+      "loss": 0.0201,
+      "step": 13086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9943534203462002e-05,
+      "loss": 2.0043,
+      "step": 13088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9938905859483478e-05,
+      "loss": 0.7776,
+      "step": 13090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9934277515504953e-05,
+      "loss": 0.9837,
+      "step": 13092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.992964917152643e-05,
+      "loss": 3.5779,
+      "step": 13094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9925020827547904e-05,
+      "loss": 0.1374,
+      "step": 13096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.992039248356938e-05,
+      "loss": 2.7978,
+      "step": 13098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9915764139590855e-05,
+      "loss": 3.157,
+      "step": 13100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.991113579561233e-05,
+      "loss": 5.375,
+      "step": 13102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9906507451633806e-05,
+      "loss": 1.2498,
+      "step": 13104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990187910765528e-05,
+      "loss": 0.6471,
+      "step": 13106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989725076367676e-05,
+      "loss": 6.4151,
+      "step": 13108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892622419698236e-05,
+      "loss": 1.0298,
+      "step": 13110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988799407571971e-05,
+      "loss": 8.0229,
+      "step": 13112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883365731741186e-05,
+      "loss": 1.5602,
+      "step": 13114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.987873738776266e-05,
+      "loss": 2.525,
+      "step": 13116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9874109043784134e-05,
+      "loss": 0.4019,
+      "step": 13118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.986948069980561e-05,
+      "loss": 2.3709,
+      "step": 13120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9864852355827085e-05,
+      "loss": 2.9419,
+      "step": 13122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.986022401184856e-05,
+      "loss": 4.3693,
+      "step": 13124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9855595667870036e-05,
+      "loss": 2.1228,
+      "step": 13126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.985096732389151e-05,
+      "loss": 2.6704,
+      "step": 13128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9846338979912986e-05,
+      "loss": 1.2145,
+      "step": 13130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9841710635934462e-05,
+      "loss": 0.0025,
+      "step": 13132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9837082291955937e-05,
+      "loss": 2.9182,
+      "step": 13134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9832453947977413e-05,
+      "loss": 1.815,
+      "step": 13136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9827825603998888e-05,
+      "loss": 2.299,
+      "step": 13138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9823197260020367e-05,
+      "loss": 1.7682,
+      "step": 13140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9818568916041842e-05,
+      "loss": 0.5695,
+      "step": 13142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9813940572063318e-05,
+      "loss": 0.8733,
+      "step": 13144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9809312228084793e-05,
+      "loss": 3.3152,
+      "step": 13146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.980468388410627e-05,
+      "loss": 0.1739,
+      "step": 13148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9800055540127744e-05,
+      "loss": 0.0021,
+      "step": 13150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.979542719614922e-05,
+      "loss": 1.3449,
+      "step": 13152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9790798852170695e-05,
+      "loss": 0.0807,
+      "step": 13154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978617050819217e-05,
+      "loss": 3.2318,
+      "step": 13156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9781542164213646e-05,
+      "loss": 1.1967,
+      "step": 13158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.977691382023512e-05,
+      "loss": 1.7287,
+      "step": 13160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9772285476256597e-05,
+      "loss": 2.5766,
+      "step": 13162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9767657132278072e-05,
+      "loss": 5.8898,
+      "step": 13164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9763028788299547e-05,
+      "loss": 2.5702,
+      "step": 13166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9758400444321023e-05,
+      "loss": 1.5105,
+      "step": 13168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9753772100342498e-05,
+      "loss": 0.5421,
+      "step": 13170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9749143756363974e-05,
+      "loss": 0.0004,
+      "step": 13172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.974451541238545e-05,
+      "loss": 0.9888,
+      "step": 13174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9739887068406925e-05,
+      "loss": 1.0898,
+      "step": 13176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.97352587244284e-05,
+      "loss": 2.1534,
+      "step": 13178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9730630380449875e-05,
+      "loss": 2.3472,
+      "step": 13180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9726002036471354e-05,
+      "loss": 4.6581,
+      "step": 13182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.972137369249283e-05,
+      "loss": 0.0004,
+      "step": 13184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9716745348514305e-05,
+      "loss": 2.2599,
+      "step": 13186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9712117004535777e-05,
+      "loss": 3.3159,
+      "step": 13188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9707488660557253e-05,
+      "loss": 1.3744,
+      "step": 13190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9702860316578728e-05,
+      "loss": 3.7766,
+      "step": 13192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9698231972600203e-05,
+      "loss": 4.8249,
+      "step": 13194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.969360362862168e-05,
+      "loss": 0.0013,
+      "step": 13196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9688975284643154e-05,
+      "loss": 1.8392,
+      "step": 13198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.968434694066463e-05,
+      "loss": 1.8035,
+      "step": 13200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9679718596686105e-05,
+      "loss": 0.0012,
+      "step": 13202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.967509025270758e-05,
+      "loss": 0.8788,
+      "step": 13204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9670461908729056e-05,
+      "loss": 2.1079,
+      "step": 13206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.966583356475053e-05,
+      "loss": 1.9554,
+      "step": 13208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9661205220772007e-05,
+      "loss": 6.6322,
+      "step": 13210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9656576876793482e-05,
+      "loss": 4.0344,
+      "step": 13212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.965194853281496e-05,
+      "loss": 0.8296,
+      "step": 13214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9647320188836436e-05,
+      "loss": 2.8819,
+      "step": 13216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9642691844857912e-05,
+      "loss": 0.4434,
+      "step": 13218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9638063500879387e-05,
+      "loss": 2.9619,
+      "step": 13220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9633435156900863e-05,
+      "loss": 1.8469,
+      "step": 13222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9628806812922338e-05,
+      "loss": 0.0019,
+      "step": 13224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9624178468943814e-05,
+      "loss": 0.7726,
+      "step": 13226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961955012496529e-05,
+      "loss": 1.4185,
+      "step": 13228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9614921780986764e-05,
+      "loss": 1.9739,
+      "step": 13230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961029343700824e-05,
+      "loss": 1.8834,
+      "step": 13232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9605665093029715e-05,
+      "loss": 0.2679,
+      "step": 13234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960103674905119e-05,
+      "loss": 1.5547,
+      "step": 13236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9596408405072666e-05,
+      "loss": 5.7028,
+      "step": 13238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.959178006109414e-05,
+      "loss": 4.1657,
+      "step": 13240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9587151717115617e-05,
+      "loss": 0.2202,
+      "step": 13242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9582523373137092e-05,
+      "loss": 2.4301,
+      "step": 13244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9577895029158568e-05,
+      "loss": 0.0003,
+      "step": 13246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9573266685180043e-05,
+      "loss": 0.6021,
+      "step": 13248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956863834120152e-05,
+      "loss": 0.7767,
+      "step": 13250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9564009997222994e-05,
+      "loss": 1.0664,
+      "step": 13252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9559381653244473e-05,
+      "loss": 2.0602,
+      "step": 13254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.955475330926595e-05,
+      "loss": 0.3002,
+      "step": 13256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9550124965287424e-05,
+      "loss": 1.0303,
+      "step": 13258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9545496621308896e-05,
+      "loss": 1.4251,
+      "step": 13260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.954086827733037e-05,
+      "loss": 0.3661,
+      "step": 13262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9536239933351847e-05,
+      "loss": 1.893,
+      "step": 13264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9531611589373322e-05,
+      "loss": 0.1163,
+      "step": 13266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9526983245394797e-05,
+      "loss": 2.4839,
+      "step": 13268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9522354901416273e-05,
+      "loss": 1.1198,
+      "step": 13270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.951772655743775e-05,
+      "loss": 0.2538,
+      "step": 13272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9513098213459224e-05,
+      "loss": 7.7177,
+      "step": 13274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.95084698694807e-05,
+      "loss": 3.3919,
+      "step": 13276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9503841525502175e-05,
+      "loss": 7.5268,
+      "step": 13278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.949921318152365e-05,
+      "loss": 1.0745,
+      "step": 13280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9494584837545125e-05,
+      "loss": 7.178,
+      "step": 13282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.94899564935666e-05,
+      "loss": 0.7392,
+      "step": 13284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9485328149588076e-05,
+      "loss": 1.693,
+      "step": 13286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9480699805609555e-05,
+      "loss": 1.3612,
+      "step": 13288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.947607146163103e-05,
+      "loss": 3.6723,
+      "step": 13290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9471443117652506e-05,
+      "loss": 2.4499,
+      "step": 13292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.946681477367398e-05,
+      "loss": 0.8863,
+      "step": 13294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9462186429695457e-05,
+      "loss": 0.0016,
+      "step": 13296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9457558085716932e-05,
+      "loss": 0.0405,
+      "step": 13298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9452929741738408e-05,
+      "loss": 1.252,
+      "step": 13300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9448301397759883e-05,
+      "loss": 0.0062,
+      "step": 13302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.944367305378136e-05,
+      "loss": 1.1732,
+      "step": 13304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9439044709802834e-05,
+      "loss": 2.16,
+      "step": 13306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.943441636582431e-05,
+      "loss": 0.651,
+      "step": 13308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9429788021845785e-05,
+      "loss": 0.7616,
+      "step": 13310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.942515967786726e-05,
+      "loss": 0.1402,
+      "step": 13312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9420531333888736e-05,
+      "loss": 0.2991,
+      "step": 13314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.941590298991021e-05,
+      "loss": 2.7966,
+      "step": 13316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9411274645931686e-05,
+      "loss": 0.0088,
+      "step": 13318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9406646301953162e-05,
+      "loss": 2.9461,
+      "step": 13320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9402017957974637e-05,
+      "loss": 1.5902,
+      "step": 13322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9397389613996113e-05,
+      "loss": 0.0095,
+      "step": 13324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9392761270017588e-05,
+      "loss": 7.1449,
+      "step": 13326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9388132926039067e-05,
+      "loss": 1.391,
+      "step": 13328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9383504582060542e-05,
+      "loss": 0.5612,
+      "step": 13330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9378876238082014e-05,
+      "loss": 1.9488,
+      "step": 13332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.937424789410349e-05,
+      "loss": 1.062,
+      "step": 13334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9369619550124965e-05,
+      "loss": 2.2017,
+      "step": 13336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.936499120614644e-05,
+      "loss": 0.2225,
+      "step": 13338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9360362862167916e-05,
+      "loss": 3.2937,
+      "step": 13340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.935573451818939e-05,
+      "loss": 4.6619,
+      "step": 13342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9351106174210867e-05,
+      "loss": 0.1726,
+      "step": 13344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9346477830232342e-05,
+      "loss": 5.0939,
+      "step": 13346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9341849486253818e-05,
+      "loss": 1.8783,
+      "step": 13348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9337221142275293e-05,
+      "loss": 2.4307,
+      "step": 13350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.933259279829677e-05,
+      "loss": 1.1087,
+      "step": 13352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9327964454318244e-05,
+      "loss": 0.0896,
+      "step": 13354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.932333611033972e-05,
+      "loss": 1.1433,
+      "step": 13356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9318707766361195e-05,
+      "loss": 0.8672,
+      "step": 13358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314079422382674e-05,
+      "loss": 1.9787,
+      "step": 13360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930945107840415e-05,
+      "loss": 2.3471,
+      "step": 13362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304822734425625e-05,
+      "loss": 4.8072,
+      "step": 13364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93001943904471e-05,
+      "loss": 1.6637,
+      "step": 13366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295566046468575e-05,
+      "loss": 6.2773,
+      "step": 13368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929093770249005e-05,
+      "loss": 0.066,
+      "step": 13370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286309358511526e-05,
+      "loss": 0.5475,
+      "step": 13372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281681014533002e-05,
+      "loss": 2.0016,
+      "step": 13374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277052670554477e-05,
+      "loss": 0.0011,
+      "step": 13376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272424326575953e-05,
+      "loss": 1.8281,
+      "step": 13378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267795982597428e-05,
+      "loss": 0.7233,
+      "step": 13380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263167638618903e-05,
+      "loss": 0.2477,
+      "step": 13382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925853929464038e-05,
+      "loss": 0.732,
+      "step": 13384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253910950661854e-05,
+      "loss": 3.6406,
+      "step": 13386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924928260668333e-05,
+      "loss": 1.425,
+      "step": 13388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244654262704805e-05,
+      "loss": 0.8571,
+      "step": 13390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924002591872628e-05,
+      "loss": 0.4945,
+      "step": 13392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235397574747756e-05,
+      "loss": 2.2872,
+      "step": 13394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 3.4967,
+      "step": 13396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226140886790707e-05,
+      "loss": 0.0177,
+      "step": 13398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221512542812182e-05,
+      "loss": 8.7932,
+      "step": 13400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921688419883366e-05,
+      "loss": 0.064,
+      "step": 13402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212255854855137e-05,
+      "loss": 3.9577,
+      "step": 13404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920762751087661e-05,
+      "loss": 1.807,
+      "step": 13406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202999166898084e-05,
+      "loss": 1.4847,
+      "step": 13408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919837082291956e-05,
+      "loss": 1.5291,
+      "step": 13410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193742478941035e-05,
+      "loss": 0.3745,
+      "step": 13412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.918911413496251e-05,
+      "loss": 1.5542,
+      "step": 13414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9184485790983986e-05,
+      "loss": 3.0253,
+      "step": 13416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.917985744700546e-05,
+      "loss": 2.1434,
+      "step": 13418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9175229103026937e-05,
+      "loss": 0.0156,
+      "step": 13420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9170600759048412e-05,
+      "loss": 1.7343,
+      "step": 13422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9165972415069887e-05,
+      "loss": 2.0155,
+      "step": 13424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9161344071091363e-05,
+      "loss": 1.801,
+      "step": 13426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9156715727112838e-05,
+      "loss": 3.9315,
+      "step": 13428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9152087383134314e-05,
+      "loss": 6.218,
+      "step": 13430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.914745903915579e-05,
+      "loss": 2.6268,
+      "step": 13432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9142830695177268e-05,
+      "loss": 1.003,
+      "step": 13434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9138202351198743e-05,
+      "loss": 3.4408,
+      "step": 13436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.913357400722022e-05,
+      "loss": 0.0215,
+      "step": 13438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9128945663241694e-05,
+      "loss": 3.2655,
+      "step": 13440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.912431731926317e-05,
+      "loss": 2.3826,
+      "step": 13442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9119688975284645e-05,
+      "loss": 1.793,
+      "step": 13444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.911506063130612e-05,
+      "loss": 3.7872,
+      "step": 13446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9110432287327596e-05,
+      "loss": 0.8993,
+      "step": 13448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.910580394334907e-05,
+      "loss": 1.6614,
+      "step": 13450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9101175599370547e-05,
+      "loss": 0.0035,
+      "step": 13452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9096547255392022e-05,
+      "loss": 1.6602,
+      "step": 13454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9091918911413498e-05,
+      "loss": 2.0293,
+      "step": 13456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087290567434973e-05,
+      "loss": 2.1621,
+      "step": 13458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908266222345645e-05,
+      "loss": 0.5429,
+      "step": 13460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078033879477924e-05,
+      "loss": 2.2707,
+      "step": 13462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90734055354994e-05,
+      "loss": 1.7344,
+      "step": 13464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068777191520875e-05,
+      "loss": 2.8464,
+      "step": 13466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906414884754235e-05,
+      "loss": 2.0617,
+      "step": 13468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9059520503563826e-05,
+      "loss": 1.9911,
+      "step": 13470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90548921595853e-05,
+      "loss": 4.3424,
+      "step": 13472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905026381560678e-05,
+      "loss": 0.7343,
+      "step": 13474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9045635471628255e-05,
+      "loss": 0.3217,
+      "step": 13476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9041007127649727e-05,
+      "loss": 3.6979,
+      "step": 13478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9036378783671203e-05,
+      "loss": 0.0049,
+      "step": 13480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9031750439692678e-05,
+      "loss": 2.2857,
+      "step": 13482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9027122095714153e-05,
+      "loss": 1.3107,
+      "step": 13484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.902249375173563e-05,
+      "loss": 1.7359,
+      "step": 13486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9017865407757104e-05,
+      "loss": 3.3044,
+      "step": 13488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.901323706377858e-05,
+      "loss": 0.3497,
+      "step": 13490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9008608719800055e-05,
+      "loss": 5.6237,
+      "step": 13492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.900398037582153e-05,
+      "loss": 1.2943,
+      "step": 13494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8999352031843006e-05,
+      "loss": 0.4622,
+      "step": 13496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.899472368786448e-05,
+      "loss": 0.1548,
+      "step": 13498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8990095343885957e-05,
+      "loss": 0.4961,
+      "step": 13500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8985466999907432e-05,
+      "loss": 2.8406,
+      "step": 13502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8980838655928908e-05,
+      "loss": 0.3669,
+      "step": 13504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8976210311950383e-05,
+      "loss": 7.0224,
+      "step": 13506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8971581967971862e-05,
+      "loss": 2.6181,
+      "step": 13508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8966953623993337e-05,
+      "loss": 2.2659,
+      "step": 13510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8962325280014813e-05,
+      "loss": 0.9308,
+      "step": 13512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8957696936036288e-05,
+      "loss": 1.0779,
+      "step": 13514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8953068592057764e-05,
+      "loss": 3.317,
+      "step": 13516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.894844024807924e-05,
+      "loss": 2.8721,
+      "step": 13518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8943811904100715e-05,
+      "loss": 3.4025,
+      "step": 13520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.893918356012219e-05,
+      "loss": 0.0011,
+      "step": 13522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8934555216143665e-05,
+      "loss": 1.824,
+      "step": 13524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.892992687216514e-05,
+      "loss": 2.6945,
+      "step": 13526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8925298528186616e-05,
+      "loss": 4.9413,
+      "step": 13528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.892067018420809e-05,
+      "loss": 3.023,
+      "step": 13530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8916041840229567e-05,
+      "loss": 3.4314,
+      "step": 13532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8911413496251043e-05,
+      "loss": 1.2994,
+      "step": 13534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8906785152272518e-05,
+      "loss": 0.0043,
+      "step": 13536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8902156808293993e-05,
+      "loss": 1.3918,
+      "step": 13538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889752846431547e-05,
+      "loss": 1.8283,
+      "step": 13540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8892900120336944e-05,
+      "loss": 0.1822,
+      "step": 13542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888827177635842e-05,
+      "loss": 1.0366,
+      "step": 13544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8883643432379895e-05,
+      "loss": 0.4983,
+      "step": 13546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8879015088401374e-05,
+      "loss": 0.2731,
+      "step": 13548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8874386744422846e-05,
+      "loss": 1.3759,
+      "step": 13550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886975840044432e-05,
+      "loss": 2.325,
+      "step": 13552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8865130056465797e-05,
+      "loss": 0.8758,
+      "step": 13554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8860501712487272e-05,
+      "loss": 3.4189,
+      "step": 13556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8855873368508748e-05,
+      "loss": 2.2829,
+      "step": 13558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8851245024530223e-05,
+      "loss": 0.2938,
+      "step": 13560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88466166805517e-05,
+      "loss": 3.2978,
+      "step": 13562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8841988336573174e-05,
+      "loss": 2.121,
+      "step": 13564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883735999259465e-05,
+      "loss": 0.0033,
+      "step": 13566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832731648616125e-05,
+      "loss": 1.3646,
+      "step": 13568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88281033046376e-05,
+      "loss": 1.1591,
+      "step": 13570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823474960659076e-05,
+      "loss": 1.3265,
+      "step": 13572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881884661668055e-05,
+      "loss": 1.2844,
+      "step": 13574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8814218272702026e-05,
+      "loss": 0.7665,
+      "step": 13576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809589928723502e-05,
+      "loss": 4.886,
+      "step": 13578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8804961584744977e-05,
+      "loss": 2.2057,
+      "step": 13580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8800333240766456e-05,
+      "loss": 0.6421,
+      "step": 13582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879570489678793e-05,
+      "loss": 5.9533,
+      "step": 13584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8791076552809407e-05,
+      "loss": 2.1755,
+      "step": 13586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8786448208830882e-05,
+      "loss": 3.3844,
+      "step": 13588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8781819864852358e-05,
+      "loss": 0.0006,
+      "step": 13590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777191520873833e-05,
+      "loss": 0.6003,
+      "step": 13592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877256317689531e-05,
+      "loss": 1.5959,
+      "step": 13594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8767934832916784e-05,
+      "loss": 3.9595,
+      "step": 13596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876330648893826e-05,
+      "loss": 0.1259,
+      "step": 13598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8758678144959735e-05,
+      "loss": 0.0061,
+      "step": 13600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.875404980098121e-05,
+      "loss": 0.9808,
+      "step": 13602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8749421457002686e-05,
+      "loss": 9.5234,
+      "step": 13604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874479311302416e-05,
+      "loss": 0.6039,
+      "step": 13606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8740164769045637e-05,
+      "loss": 2.8311,
+      "step": 13608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8735536425067112e-05,
+      "loss": 0.4645,
+      "step": 13610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8730908081088587e-05,
+      "loss": 2.3792,
+      "step": 13612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8726279737110063e-05,
+      "loss": 0.0067,
+      "step": 13614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8721651393131538e-05,
+      "loss": 1.2012,
+      "step": 13616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8717023049153014e-05,
+      "loss": 2.3347,
+      "step": 13618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871239470517449e-05,
+      "loss": 4.5485,
+      "step": 13620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8707766361195965e-05,
+      "loss": 1.3667,
+      "step": 13622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870313801721744e-05,
+      "loss": 0.704,
+      "step": 13624
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8698509673238915e-05,
+      "loss": 4.1856,
+      "step": 13626
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869388132926039e-05,
+      "loss": 4.8153,
+      "step": 13628
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8689252985281866e-05,
+      "loss": 2.9679,
+      "step": 13630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8684624641303342e-05,
+      "loss": 3.6085,
+      "step": 13632
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8679996297324817e-05,
+      "loss": 0.2644,
+      "step": 13634
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8675367953346293e-05,
+      "loss": 2.8538,
+      "step": 13636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8670739609367768e-05,
+      "loss": 0.0032,
+      "step": 13638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8666111265389243e-05,
+      "loss": 0.402,
+      "step": 13640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.866148292141072e-05,
+      "loss": 0.2989,
+      "step": 13642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8656854577432194e-05,
+      "loss": 0.0036,
+      "step": 13644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.865222623345367e-05,
+      "loss": 2.5014,
+      "step": 13646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8647597889475145e-05,
+      "loss": 0.714,
+      "step": 13648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.864296954549662e-05,
+      "loss": 3.6613,
+      "step": 13650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8638341201518096e-05,
+      "loss": 0.1582,
+      "step": 13652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8633712857539575e-05,
+      "loss": 0.006,
+      "step": 13654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.862908451356105e-05,
+      "loss": 0.0009,
+      "step": 13656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8624456169582526e-05,
+      "loss": 2.0837,
+      "step": 13658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8619827825604e-05,
+      "loss": 0.0553,
+      "step": 13660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8615199481625476e-05,
+      "loss": 0.1025,
+      "step": 13662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8610571137646952e-05,
+      "loss": 0.8116,
+      "step": 13664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8605942793668427e-05,
+      "loss": 0.0281,
+      "step": 13666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8601314449689903e-05,
+      "loss": 5.2014,
+      "step": 13668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8596686105711378e-05,
+      "loss": 1.9461,
+      "step": 13670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8592057761732854e-05,
+      "loss": 0.0005,
+      "step": 13672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.858742941775433e-05,
+      "loss": 3.2973,
+      "step": 13674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8582801073775804e-05,
+      "loss": 4.205,
+      "step": 13676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.857817272979728e-05,
+      "loss": 3.4792,
+      "step": 13678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8573544385818755e-05,
+      "loss": 2.1713,
+      "step": 13680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.856891604184023e-05,
+      "loss": 0.0022,
+      "step": 13682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8564287697861706e-05,
+      "loss": 2.7074,
+      "step": 13684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.855965935388318e-05,
+      "loss": 4.5783,
+      "step": 13686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8555031009904657e-05,
+      "loss": 2.8419,
+      "step": 13688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8550402665926132e-05,
+      "loss": 4.0734,
+      "step": 13690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8545774321947608e-05,
+      "loss": 2.8184,
+      "step": 13692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8541145977969083e-05,
+      "loss": 9.5152,
+      "step": 13694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.853651763399056e-05,
+      "loss": 0.5262,
+      "step": 13696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8531889290012034e-05,
+      "loss": 1.9495,
+      "step": 13698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.852726094603351e-05,
+      "loss": 5.6337,
+      "step": 13700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8522632602054985e-05,
+      "loss": 6.1846,
+      "step": 13702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.851800425807646e-05,
+      "loss": 4.0793,
+      "step": 13704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8513375914097936e-05,
+      "loss": 0.2953,
+      "step": 13706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.850874757011941e-05,
+      "loss": 1.2353,
+      "step": 13708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8504119226140887e-05,
+      "loss": 0.1664,
+      "step": 13710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8499490882162362e-05,
+      "loss": 1.1889,
+      "step": 13712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8494862538183837e-05,
+      "loss": 0.004,
+      "step": 13714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8490234194205313e-05,
+      "loss": 1.6534,
+      "step": 13716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.848560585022679e-05,
+      "loss": 0.355,
+      "step": 13718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8480977506248264e-05,
+      "loss": 1.8386,
+      "step": 13720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.847634916226974e-05,
+      "loss": 0.0067,
+      "step": 13722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8471720818291215e-05,
+      "loss": 0.0006,
+      "step": 13724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.846709247431269e-05,
+      "loss": 0.096,
+      "step": 13726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.846246413033417e-05,
+      "loss": 1.469,
+      "step": 13728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8457835786355644e-05,
+      "loss": 0.0005,
+      "step": 13730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.845320744237712e-05,
+      "loss": 3.3645,
+      "step": 13732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8448579098398595e-05,
+      "loss": 1.5309,
+      "step": 13734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.844395075442007e-05,
+      "loss": 3.5192,
+      "step": 13736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8439322410441546e-05,
+      "loss": 1.0765,
+      "step": 13738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.843469406646302e-05,
+      "loss": 2.5575,
+      "step": 13740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8430065722484497e-05,
+      "loss": 3.3089,
+      "step": 13742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8425437378505972e-05,
+      "loss": 1.5296,
+      "step": 13744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8420809034527448e-05,
+      "loss": 0.5082,
+      "step": 13746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8416180690548923e-05,
+      "loss": 8.9516,
+      "step": 13748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.84115523465704e-05,
+      "loss": 1.272,
+      "step": 13750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8406924002591874e-05,
+      "loss": 2.1876,
+      "step": 13752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.840229565861335e-05,
+      "loss": 3.1165,
+      "step": 13754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8397667314634825e-05,
+      "loss": 2.5071,
+      "step": 13756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.83930389706563e-05,
+      "loss": 0.8418,
+      "step": 13758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8388410626677776e-05,
+      "loss": 2.0806,
+      "step": 13760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.838378228269925e-05,
+      "loss": 1.7485,
+      "step": 13762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8379153938720726e-05,
+      "loss": 0.0035,
+      "step": 13764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8374525594742202e-05,
+      "loss": 0.5644,
+      "step": 13766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8369897250763677e-05,
+      "loss": 1.09,
+      "step": 13768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8365268906785153e-05,
+      "loss": 0.0922,
+      "step": 13770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8360640562806628e-05,
+      "loss": 0.0333,
+      "step": 13772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8356012218828104e-05,
+      "loss": 0.5531,
+      "step": 13774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.835138387484958e-05,
+      "loss": 0.5936,
+      "step": 13776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8346755530871054e-05,
+      "loss": 1.3969,
+      "step": 13778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.834212718689253e-05,
+      "loss": 4.8334,
+      "step": 13780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8337498842914005e-05,
+      "loss": 1.6664,
+      "step": 13782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.833287049893548e-05,
+      "loss": 2.933,
+      "step": 13784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8328242154956956e-05,
+      "loss": 0.5058,
+      "step": 13786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.832361381097843e-05,
+      "loss": 2.9082,
+      "step": 13788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8318985466999907e-05,
+      "loss": 0.6259,
+      "step": 13790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8314357123021382e-05,
+      "loss": 0.0009,
+      "step": 13792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8309728779042858e-05,
+      "loss": 0.7306,
+      "step": 13794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8305100435064333e-05,
+      "loss": 0.128,
+      "step": 13796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.830047209108581e-05,
+      "loss": 0.1012,
+      "step": 13798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8295843747107284e-05,
+      "loss": 2.7364,
+      "step": 13800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8291215403128763e-05,
+      "loss": 8.0037,
+      "step": 13802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.828658705915024e-05,
+      "loss": 0.38,
+      "step": 13804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8281958715171714e-05,
+      "loss": 4.2929,
+      "step": 13806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.827733037119319e-05,
+      "loss": 2.4929,
+      "step": 13808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8272702027214665e-05,
+      "loss": 0.0006,
+      "step": 13810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.826807368323614e-05,
+      "loss": 2.191,
+      "step": 13812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8263445339257615e-05,
+      "loss": 0.9902,
+      "step": 13814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.825881699527909e-05,
+      "loss": 0.0587,
+      "step": 13816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8254188651300566e-05,
+      "loss": 1.1723,
+      "step": 13818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8249560307322042e-05,
+      "loss": 0.0024,
+      "step": 13820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8244931963343517e-05,
+      "loss": 2.1411,
+      "step": 13822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8240303619364993e-05,
+      "loss": 1.1483,
+      "step": 13824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8235675275386468e-05,
+      "loss": 0.679,
+      "step": 13826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8231046931407943e-05,
+      "loss": 5.6413,
+      "step": 13828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.822641858742942e-05,
+      "loss": 1.399,
+      "step": 13830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8221790243450894e-05,
+      "loss": 1.1885,
+      "step": 13832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.821716189947237e-05,
+      "loss": 2.2914,
+      "step": 13834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8212533555493845e-05,
+      "loss": 0.331,
+      "step": 13836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.820790521151532e-05,
+      "loss": 1.0843,
+      "step": 13838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8203276867536796e-05,
+      "loss": 0.7516,
+      "step": 13840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.819864852355827e-05,
+      "loss": 3.0404,
+      "step": 13842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8194020179579747e-05,
+      "loss": 4.6834,
+      "step": 13844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8189391835601222e-05,
+      "loss": 0.3035,
+      "step": 13846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8184763491622698e-05,
+      "loss": 0.0032,
+      "step": 13848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8180135147644173e-05,
+      "loss": 2.2313,
+      "step": 13850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.817550680366565e-05,
+      "loss": 0.031,
+      "step": 13852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8170878459687124e-05,
+      "loss": 3.0287,
+      "step": 13854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.81662501157086e-05,
+      "loss": 0.1104,
+      "step": 13856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8161621771730075e-05,
+      "loss": 0.3485,
+      "step": 13858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.815699342775155e-05,
+      "loss": 0.5355,
+      "step": 13860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8152365083773026e-05,
+      "loss": 3.9581,
+      "step": 13862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.81477367397945e-05,
+      "loss": 0.0016,
+      "step": 13864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8143108395815977e-05,
+      "loss": 0.0009,
+      "step": 13866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8138480051837452e-05,
+      "loss": 0.6097,
+      "step": 13868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8133851707858927e-05,
+      "loss": 0.526,
+      "step": 13870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8129223363880403e-05,
+      "loss": 5.0417,
+      "step": 13872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8124595019901878e-05,
+      "loss": 0.0045,
+      "step": 13874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8119966675923357e-05,
+      "loss": 2.9541,
+      "step": 13876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8115338331944832e-05,
+      "loss": 6.5442,
+      "step": 13878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8110709987966308e-05,
+      "loss": 3.7948,
+      "step": 13880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8106081643987783e-05,
+      "loss": 0.4933,
+      "step": 13882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.810145330000926e-05,
+      "loss": 0.227,
+      "step": 13884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8096824956030734e-05,
+      "loss": 2.2246,
+      "step": 13886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.809219661205221e-05,
+      "loss": 8.2544,
+      "step": 13888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8087568268073685e-05,
+      "loss": 4.5197,
+      "step": 13890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.808293992409516e-05,
+      "loss": 0.3681,
+      "step": 13892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8078311580116636e-05,
+      "loss": 0.4646,
+      "step": 13894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.807368323613811e-05,
+      "loss": 0.6976,
+      "step": 13896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8069054892159587e-05,
+      "loss": 3.012,
+      "step": 13898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8064426548181062e-05,
+      "loss": 2.4573,
+      "step": 13900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8059798204202538e-05,
+      "loss": 0.0005,
+      "step": 13902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8055169860224013e-05,
+      "loss": 0.0065,
+      "step": 13904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.805054151624549e-05,
+      "loss": 1.3884,
+      "step": 13906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8045913172266964e-05,
+      "loss": 3.2777,
+      "step": 13908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.804128482828844e-05,
+      "loss": 0.4445,
+      "step": 13910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8036656484309915e-05,
+      "loss": 0.008,
+      "step": 13912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.803202814033139e-05,
+      "loss": 0.0562,
+      "step": 13914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8027399796352866e-05,
+      "loss": 3.1283,
+      "step": 13916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.802277145237434e-05,
+      "loss": 0.1598,
+      "step": 13918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8018143108395816e-05,
+      "loss": 0.7442,
+      "step": 13920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8013514764417292e-05,
+      "loss": 0.7279,
+      "step": 13922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8008886420438767e-05,
+      "loss": 5.597,
+      "step": 13924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8004258076460243e-05,
+      "loss": 2.4086,
+      "step": 13926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7999629732481718e-05,
+      "loss": 1.517,
+      "step": 13928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7995001388503194e-05,
+      "loss": 0.0006,
+      "step": 13930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.799037304452467e-05,
+      "loss": 0.0005,
+      "step": 13932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7985744700546144e-05,
+      "loss": 4.8246,
+      "step": 13934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.798111635656762e-05,
+      "loss": 2.6006,
+      "step": 13936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7976488012589095e-05,
+      "loss": 0.1007,
+      "step": 13938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.797185966861057e-05,
+      "loss": 1.4297,
+      "step": 13940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7967231324632046e-05,
+      "loss": 0.1441,
+      "step": 13942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.796260298065352e-05,
+      "loss": 1.8149,
+      "step": 13944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7957974636674997e-05,
+      "loss": 1.6269,
+      "step": 13946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7953346292696476e-05,
+      "loss": 0.0004,
+      "step": 13948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 2.3885,
+      "step": 13950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7944089604739427e-05,
+      "loss": 1.1743,
+      "step": 13952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7939461260760902e-05,
+      "loss": 3.4835,
+      "step": 13954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7934832916782377e-05,
+      "loss": 4.7977,
+      "step": 13956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7930204572803853e-05,
+      "loss": 1.7986,
+      "step": 13958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7925576228825328e-05,
+      "loss": 2.5674,
+      "step": 13960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7920947884846804e-05,
+      "loss": 0.0008,
+      "step": 13962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.791631954086828e-05,
+      "loss": 0.0019,
+      "step": 13964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7911691196889755e-05,
+      "loss": 1.8908,
+      "step": 13966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.790706285291123e-05,
+      "loss": 4.4968,
+      "step": 13968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7902434508932705e-05,
+      "loss": 2.6313,
+      "step": 13970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.789780616495418e-05,
+      "loss": 4.3402,
+      "step": 13972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7893177820975656e-05,
+      "loss": 1.689,
+      "step": 13974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.788854947699713e-05,
+      "loss": 3.3673,
+      "step": 13976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7883921133018607e-05,
+      "loss": 3.8666,
+      "step": 13978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7879292789040083e-05,
+      "loss": 0.0342,
+      "step": 13980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7874664445061558e-05,
+      "loss": 5.9796,
+      "step": 13982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7870036101083033e-05,
+      "loss": 3.1522,
+      "step": 13984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.786540775710451e-05,
+      "loss": 3.4608,
+      "step": 13986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7860779413125984e-05,
+      "loss": 2.4209,
+      "step": 13988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.785615106914746e-05,
+      "loss": 0.0036,
+      "step": 13990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7851522725168935e-05,
+      "loss": 0.084,
+      "step": 13992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.784689438119041e-05,
+      "loss": 0.273,
+      "step": 13994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7842266037211886e-05,
+      "loss": 1.0989,
+      "step": 13996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.783763769323336e-05,
+      "loss": 0.5135,
+      "step": 13998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7833009349254837e-05,
+      "loss": 5.2966,
+      "step": 14000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7828381005276312e-05,
+      "loss": 1.5025,
+      "step": 14002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7823752661297788e-05,
+      "loss": 2.2765,
+      "step": 14004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7819124317319263e-05,
+      "loss": 3.2422,
+      "step": 14006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.781449597334074e-05,
+      "loss": 1.2087,
+      "step": 14008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7809867629362214e-05,
+      "loss": 5.4052,
+      "step": 14010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.780523928538369e-05,
+      "loss": 2.9954,
+      "step": 14012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7800610941405165e-05,
+      "loss": 2.333,
+      "step": 14014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.779598259742664e-05,
+      "loss": 3.5513,
+      "step": 14016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7791354253448116e-05,
+      "loss": 0.834,
+      "step": 14018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.778672590946959e-05,
+      "loss": 0.5189,
+      "step": 14020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.778209756549107e-05,
+      "loss": 0.9156,
+      "step": 14022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7777469221512545e-05,
+      "loss": 2.3368,
+      "step": 14024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.777284087753402e-05,
+      "loss": 2.5395,
+      "step": 14026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7768212533555496e-05,
+      "loss": 2.3215,
+      "step": 14028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.776358418957697e-05,
+      "loss": 1.3904,
+      "step": 14030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7758955845598447e-05,
+      "loss": 0.0558,
+      "step": 14032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7754327501619922e-05,
+      "loss": 0.3764,
+      "step": 14034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7749699157641398e-05,
+      "loss": 2.5968,
+      "step": 14036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7745070813662873e-05,
+      "loss": 1.8432,
+      "step": 14038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.774044246968435e-05,
+      "loss": 6.6889,
+      "step": 14040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7735814125705824e-05,
+      "loss": 2.1079,
+      "step": 14042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.77311857817273e-05,
+      "loss": 0.243,
+      "step": 14044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7726557437748775e-05,
+      "loss": 0.0754,
+      "step": 14046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.772192909377025e-05,
+      "loss": 0.0792,
+      "step": 14048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7717300749791726e-05,
+      "loss": 2.5114,
+      "step": 14050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.77126724058132e-05,
+      "loss": 2.5162,
+      "step": 14052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7708044061834677e-05,
+      "loss": 1.3254,
+      "step": 14054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7703415717856152e-05,
+      "loss": 1.4562,
+      "step": 14056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7698787373877627e-05,
+      "loss": 2.6882,
+      "step": 14058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7694159029899103e-05,
+      "loss": 4.1705,
+      "step": 14060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.768953068592058e-05,
+      "loss": 5.2712,
+      "step": 14062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7684902341942054e-05,
+      "loss": 2.4092,
+      "step": 14064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.768027399796353e-05,
+      "loss": 1.7373,
+      "step": 14066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7675645653985005e-05,
+      "loss": 4.2599,
+      "step": 14068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.767101731000648e-05,
+      "loss": 0.0008,
+      "step": 14070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7666388966027955e-05,
+      "loss": 2.8173,
+      "step": 14072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.766176062204943e-05,
+      "loss": 2.1836,
+      "step": 14074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7657132278070906e-05,
+      "loss": 1.623,
+      "step": 14076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7652503934092382e-05,
+      "loss": 3.3051,
+      "step": 14078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7647875590113857e-05,
+      "loss": 0.0016,
+      "step": 14080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7643247246135333e-05,
+      "loss": 1.4861,
+      "step": 14082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7638618902156808e-05,
+      "loss": 1.7357,
+      "step": 14084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7633990558178283e-05,
+      "loss": 1.275,
+      "step": 14086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.762936221419976e-05,
+      "loss": 0.2014,
+      "step": 14088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7624733870221234e-05,
+      "loss": 2.9736,
+      "step": 14090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.762010552624271e-05,
+      "loss": 2.7779,
+      "step": 14092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7615477182264185e-05,
+      "loss": 2.2258,
+      "step": 14094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7610848838285664e-05,
+      "loss": 1.8074,
+      "step": 14096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.760622049430714e-05,
+      "loss": 1.2776,
+      "step": 14098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7601592150328615e-05,
+      "loss": 0.0108,
+      "step": 14100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.759696380635009e-05,
+      "loss": 0.4278,
+      "step": 14102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7592335462371566e-05,
+      "loss": 0.4535,
+      "step": 14104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.758770711839304e-05,
+      "loss": 0.003,
+      "step": 14106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7583078774414516e-05,
+      "loss": 1.889,
+      "step": 14108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7578450430435992e-05,
+      "loss": 0.0005,
+      "step": 14110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7573822086457467e-05,
+      "loss": 0.3605,
+      "step": 14112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7569193742478943e-05,
+      "loss": 0.0007,
+      "step": 14114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7564565398500418e-05,
+      "loss": 1.0247,
+      "step": 14116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7559937054521894e-05,
+      "loss": 0.4573,
+      "step": 14118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.755530871054337e-05,
+      "loss": 1.6794,
+      "step": 14120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7550680366564844e-05,
+      "loss": 2.1704,
+      "step": 14122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.754605202258632e-05,
+      "loss": 4.0895,
+      "step": 14124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7541423678607795e-05,
+      "loss": 0.8793,
+      "step": 14126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.753679533462927e-05,
+      "loss": 2.5844,
+      "step": 14128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7532166990650746e-05,
+      "loss": 0.2051,
+      "step": 14130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.752753864667222e-05,
+      "loss": 2.2902,
+      "step": 14132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7522910302693697e-05,
+      "loss": 4.0179,
+      "step": 14134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7518281958715172e-05,
+      "loss": 0.0017,
+      "step": 14136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7513653614736648e-05,
+      "loss": 4.8414,
+      "step": 14138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7509025270758123e-05,
+      "loss": 2.2602,
+      "step": 14140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.75043969267796e-05,
+      "loss": 1.85,
+      "step": 14142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7499768582801074e-05,
+      "loss": 6.2589,
+      "step": 14144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.749514023882255e-05,
+      "loss": 1.1886,
+      "step": 14146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7490511894844025e-05,
+      "loss": 0.6075,
+      "step": 14148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.74858835508655e-05,
+      "loss": 0.9095,
+      "step": 14150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7481255206886976e-05,
+      "loss": 2.2343,
+      "step": 14152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.747662686290845e-05,
+      "loss": 0.4552,
+      "step": 14154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7471998518929927e-05,
+      "loss": 1.708,
+      "step": 14156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7467370174951402e-05,
+      "loss": 0.042,
+      "step": 14158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7462741830972877e-05,
+      "loss": 2.2137,
+      "step": 14160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7458113486994353e-05,
+      "loss": 3.0299,
+      "step": 14162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.745348514301583e-05,
+      "loss": 1.7795,
+      "step": 14164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7448856799037304e-05,
+      "loss": 0.8973,
+      "step": 14166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7444228455058783e-05,
+      "loss": 1.7542,
+      "step": 14168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7439600111080258e-05,
+      "loss": 0.6673,
+      "step": 14170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7434971767101733e-05,
+      "loss": 4.5049,
+      "step": 14172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.743034342312321e-05,
+      "loss": 0.1665,
+      "step": 14174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7425715079144684e-05,
+      "loss": 0.4135,
+      "step": 14176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.742108673516616e-05,
+      "loss": 1.6589,
+      "step": 14178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7416458391187635e-05,
+      "loss": 3.176,
+      "step": 14180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.741183004720911e-05,
+      "loss": 0.3027,
+      "step": 14182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7407201703230586e-05,
+      "loss": 1.937,
+      "step": 14184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.740257335925206e-05,
+      "loss": 0.9323,
+      "step": 14186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7397945015273537e-05,
+      "loss": 1.0712,
+      "step": 14188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7393316671295012e-05,
+      "loss": 5.5913,
+      "step": 14190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7388688327316488e-05,
+      "loss": 0.455,
+      "step": 14192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7384059983337963e-05,
+      "loss": 3.7109,
+      "step": 14194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.737943163935944e-05,
+      "loss": 1.6452,
+      "step": 14196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7374803295380914e-05,
+      "loss": 0.8799,
+      "step": 14198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.737017495140239e-05,
+      "loss": 2.719,
+      "step": 14200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7365546607423865e-05,
+      "loss": 1.7441,
+      "step": 14202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.736091826344534e-05,
+      "loss": 0.4166,
+      "step": 14204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7356289919466816e-05,
+      "loss": 2.2928,
+      "step": 14206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.735166157548829e-05,
+      "loss": 0.6742,
+      "step": 14208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7347033231509767e-05,
+      "loss": 3.7499,
+      "step": 14210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7342404887531242e-05,
+      "loss": 3.6805,
+      "step": 14212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7337776543552717e-05,
+      "loss": 0.1114,
+      "step": 14214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7333148199574193e-05,
+      "loss": 5.7093,
+      "step": 14216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7328519855595668e-05,
+      "loss": 2.7732,
+      "step": 14218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7323891511617144e-05,
+      "loss": 4.7145,
+      "step": 14220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.731926316763862e-05,
+      "loss": 0.4944,
+      "step": 14222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7314634823660094e-05,
+      "loss": 1.701,
+      "step": 14224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.731000647968157e-05,
+      "loss": 1.7965,
+      "step": 14226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7305378135703045e-05,
+      "loss": 0.8086,
+      "step": 14228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.730074979172452e-05,
+      "loss": 1.4546,
+      "step": 14230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7296121447745996e-05,
+      "loss": 0.0336,
+      "step": 14232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.729149310376747e-05,
+      "loss": 2.2752,
+      "step": 14234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7286864759788947e-05,
+      "loss": 0.4761,
+      "step": 14236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7282236415810422e-05,
+      "loss": 0.0008,
+      "step": 14238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7277608071831898e-05,
+      "loss": 4.2546,
+      "step": 14240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7272979727853377e-05,
+      "loss": 1.3129,
+      "step": 14242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7268351383874852e-05,
+      "loss": 3.3543,
+      "step": 14244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7263723039896328e-05,
+      "loss": 0.6315,
+      "step": 14246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7259094695917803e-05,
+      "loss": 2.9811,
+      "step": 14248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.725446635193928e-05,
+      "loss": 3.6715,
+      "step": 14250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7249838007960754e-05,
+      "loss": 0.0035,
+      "step": 14252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.724520966398223e-05,
+      "loss": 0.1769,
+      "step": 14254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7240581320003705e-05,
+      "loss": 0.0534,
+      "step": 14256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.723595297602518e-05,
+      "loss": 0.7155,
+      "step": 14258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7231324632046656e-05,
+      "loss": 3.239,
+      "step": 14260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.722669628806813e-05,
+      "loss": 5.7438,
+      "step": 14262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7222067944089606e-05,
+      "loss": 0.4151,
+      "step": 14264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7217439600111082e-05,
+      "loss": 0.0713,
+      "step": 14266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7212811256132557e-05,
+      "loss": 5.3267,
+      "step": 14268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7208182912154033e-05,
+      "loss": 1.5913,
+      "step": 14270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7203554568175508e-05,
+      "loss": 1.466,
+      "step": 14272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7198926224196983e-05,
+      "loss": 3.3951,
+      "step": 14274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.719429788021846e-05,
+      "loss": 0.2657,
+      "step": 14276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7189669536239934e-05,
+      "loss": 1.6616,
+      "step": 14278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.718504119226141e-05,
+      "loss": 3.9433,
+      "step": 14280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7180412848282885e-05,
+      "loss": 1.3582,
+      "step": 14282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.717578450430436e-05,
+      "loss": 2.9668,
+      "step": 14284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7171156160325836e-05,
+      "loss": 2.3829,
+      "step": 14286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.716652781634731e-05,
+      "loss": 1.9184,
+      "step": 14288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7161899472368787e-05,
+      "loss": 4.6674,
+      "step": 14290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7157271128390262e-05,
+      "loss": 1.0855,
+      "step": 14292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7152642784411738e-05,
+      "loss": 0.0407,
+      "step": 14294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7148014440433213e-05,
+      "loss": 1.6122,
+      "step": 14296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.714338609645469e-05,
+      "loss": 6.2775,
+      "step": 14298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7138757752476164e-05,
+      "loss": 0.0018,
+      "step": 14300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.713412940849764e-05,
+      "loss": 1.5273,
+      "step": 14302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7129501064519115e-05,
+      "loss": 0.3795,
+      "step": 14304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.712487272054059e-05,
+      "loss": 4.8135,
+      "step": 14306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7120244376562066e-05,
+      "loss": 2.4487,
+      "step": 14308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.711561603258354e-05,
+      "loss": 0.6626,
+      "step": 14310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7110987688605017e-05,
+      "loss": 0.003,
+      "step": 14312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7106359344626492e-05,
+      "loss": 3.9032,
+      "step": 14314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.710173100064797e-05,
+      "loss": 5.8713,
+      "step": 14316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7097102656669446e-05,
+      "loss": 7.7147,
+      "step": 14318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.709247431269092e-05,
+      "loss": 1.2029,
+      "step": 14320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7087845968712397e-05,
+      "loss": 0.0137,
+      "step": 14322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7083217624733872e-05,
+      "loss": 1.214,
+      "step": 14324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7078589280755348e-05,
+      "loss": 0.0131,
+      "step": 14326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7073960936776823e-05,
+      "loss": 1.149,
+      "step": 14328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.70693325927983e-05,
+      "loss": 0.0121,
+      "step": 14330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7064704248819774e-05,
+      "loss": 1.4038,
+      "step": 14332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.706007590484125e-05,
+      "loss": 0.805,
+      "step": 14334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7055447560862725e-05,
+      "loss": 0.6927,
+      "step": 14336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.70508192168842e-05,
+      "loss": 0.961,
+      "step": 14338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7046190872905676e-05,
+      "loss": 1.6342,
+      "step": 14340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.704156252892715e-05,
+      "loss": 0.0606,
+      "step": 14342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7036934184948627e-05,
+      "loss": 0.3379,
+      "step": 14344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7032305840970102e-05,
+      "loss": 0.6071,
+      "step": 14346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7027677496991578e-05,
+      "loss": 0.0979,
+      "step": 14348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7023049153013053e-05,
+      "loss": 0.1134,
+      "step": 14350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.701842080903453e-05,
+      "loss": 0.0223,
+      "step": 14352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7013792465056004e-05,
+      "loss": 0.6196,
+      "step": 14354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.700916412107748e-05,
+      "loss": 0.6496,
+      "step": 14356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7004535777098955e-05,
+      "loss": 5.8406,
+      "step": 14358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.699990743312043e-05,
+      "loss": 0.3483,
+      "step": 14360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6995279089141906e-05,
+      "loss": 0.371,
+      "step": 14362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.699065074516338e-05,
+      "loss": 0.9759,
+      "step": 14364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6986022401184856e-05,
+      "loss": 0.513,
+      "step": 14366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6981394057206332e-05,
+      "loss": 0.024,
+      "step": 14368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6976765713227807e-05,
+      "loss": 4.3694,
+      "step": 14370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6972137369249283e-05,
+      "loss": 1.6888,
+      "step": 14372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6967509025270758e-05,
+      "loss": 2.7927,
+      "step": 14374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6962880681292234e-05,
+      "loss": 0.0003,
+      "step": 14376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.695825233731371e-05,
+      "loss": 3.7456,
+      "step": 14378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6953623993335184e-05,
+      "loss": 7.3766,
+      "step": 14380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.694899564935666e-05,
+      "loss": 0.4556,
+      "step": 14382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6944367305378135e-05,
+      "loss": 4.9924,
+      "step": 14384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.693973896139961e-05,
+      "loss": 0.0059,
+      "step": 14386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6935110617421086e-05,
+      "loss": 0.637,
+      "step": 14388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6930482273442565e-05,
+      "loss": 1.8404,
+      "step": 14390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.692585392946404e-05,
+      "loss": 0.1671,
+      "step": 14392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6921225585485516e-05,
+      "loss": 2.9233,
+      "step": 14394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.691659724150699e-05,
+      "loss": 3.4532,
+      "step": 14396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6911968897528467e-05,
+      "loss": 8.511,
+      "step": 14398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6907340553549942e-05,
+      "loss": 1.9012,
+      "step": 14400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6902712209571417e-05,
+      "loss": 0.0021,
+      "step": 14402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6898083865592893e-05,
+      "loss": 2.4593,
+      "step": 14404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6893455521614368e-05,
+      "loss": 4.9771,
+      "step": 14406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6888827177635844e-05,
+      "loss": 0.687,
+      "step": 14408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.688419883365732e-05,
+      "loss": 3.7834,
+      "step": 14410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6879570489678795e-05,
+      "loss": 0.2298,
+      "step": 14412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.687494214570027e-05,
+      "loss": 0.0003,
+      "step": 14414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6870313801721745e-05,
+      "loss": 2.4668,
+      "step": 14416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.686568545774322e-05,
+      "loss": 4.7613,
+      "step": 14418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6861057113764693e-05,
+      "loss": 4.0283,
+      "step": 14420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.685642876978617e-05,
+      "loss": 0.9501,
+      "step": 14422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6851800425807647e-05,
+      "loss": 0.5037,
+      "step": 14424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6847172081829123e-05,
+      "loss": 0.4629,
+      "step": 14426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6842543737850598e-05,
+      "loss": 2.6493,
+      "step": 14428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6837915393872073e-05,
+      "loss": 0.3727,
+      "step": 14430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.683328704989355e-05,
+      "loss": 1.3007,
+      "step": 14432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6828658705915024e-05,
+      "loss": 0.5354,
+      "step": 14434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.68240303619365e-05,
+      "loss": 5.4703,
+      "step": 14436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6819402017957975e-05,
+      "loss": 0.9454,
+      "step": 14438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.681477367397945e-05,
+      "loss": 0.0902,
+      "step": 14440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6810145330000926e-05,
+      "loss": 1.3643,
+      "step": 14442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.68055169860224e-05,
+      "loss": 2.4239,
+      "step": 14444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6800888642043877e-05,
+      "loss": 1.1963,
+      "step": 14446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6796260298065352e-05,
+      "loss": 1.0935,
+      "step": 14448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6791631954086828e-05,
+      "loss": 0.4362,
+      "step": 14450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6787003610108303e-05,
+      "loss": 0.6111,
+      "step": 14452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.678237526612978e-05,
+      "loss": 0.2667,
+      "step": 14454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6777746922151254e-05,
+      "loss": 6.5233,
+      "step": 14456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.677311857817273e-05,
+      "loss": 0.4211,
+      "step": 14458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6768490234194205e-05,
+      "loss": 2.611,
+      "step": 14460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6763861890215684e-05,
+      "loss": 0.0024,
+      "step": 14462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.675923354623716e-05,
+      "loss": 0.7943,
+      "step": 14464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6754605202258634e-05,
+      "loss": 3.336,
+      "step": 14466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.674997685828011e-05,
+      "loss": 0.0063,
+      "step": 14468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6745348514301585e-05,
+      "loss": 0.4712,
+      "step": 14470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.674072017032306e-05,
+      "loss": 0.0798,
+      "step": 14472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6736091826344536e-05,
+      "loss": 1.7138,
+      "step": 14474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.673146348236601e-05,
+      "loss": 7.8274,
+      "step": 14476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6726835138387487e-05,
+      "loss": 5.034,
+      "step": 14478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6722206794408962e-05,
+      "loss": 2.2502,
+      "step": 14480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6717578450430438e-05,
+      "loss": 1.1995,
+      "step": 14482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6712950106451913e-05,
+      "loss": 0.7682,
+      "step": 14484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.670832176247339e-05,
+      "loss": 4.1127,
+      "step": 14486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6703693418494864e-05,
+      "loss": 1.9902,
+      "step": 14488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.669906507451634e-05,
+      "loss": 2.3483,
+      "step": 14490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.669443673053781e-05,
+      "loss": 1.8379,
+      "step": 14492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6689808386559287e-05,
+      "loss": 4.3483,
+      "step": 14494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6685180042580766e-05,
+      "loss": 0.0016,
+      "step": 14496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.668055169860224e-05,
+      "loss": 1.3972,
+      "step": 14498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6675923354623717e-05,
+      "loss": 0.1004,
+      "step": 14500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6671295010645192e-05,
+      "loss": 0.1341,
+      "step": 14502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.1009,
+      "step": 14504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6662038322688143e-05,
+      "loss": 6.019,
+      "step": 14506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.665740997870962e-05,
+      "loss": 3.1944,
+      "step": 14508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6652781634731094e-05,
+      "loss": 0.1878,
+      "step": 14510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.664815329075257e-05,
+      "loss": 3.0428,
+      "step": 14512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6643524946774045e-05,
+      "loss": 4.2022,
+      "step": 14514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.663889660279552e-05,
+      "loss": 1.8916,
+      "step": 14516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6634268258816995e-05,
+      "loss": 0.7287,
+      "step": 14518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.662963991483847e-05,
+      "loss": 2.567,
+      "step": 14520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6625011570859946e-05,
+      "loss": 0.0839,
+      "step": 14522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6620383226881422e-05,
+      "loss": 2.2278,
+      "step": 14524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6615754882902897e-05,
+      "loss": 0.792,
+      "step": 14526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6611126538924373e-05,
+      "loss": 2.7499,
+      "step": 14528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6606498194945848e-05,
+      "loss": 5.141,
+      "step": 14530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6601869850967323e-05,
+      "loss": 0.1278,
+      "step": 14532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.65972415069888e-05,
+      "loss": 1.784,
+      "step": 14534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6592613163010278e-05,
+      "loss": 3.2108,
+      "step": 14536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6587984819031753e-05,
+      "loss": 3.0949,
+      "step": 14538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.658335647505323e-05,
+      "loss": 2.1306,
+      "step": 14540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6578728131074704e-05,
+      "loss": 0.6708,
+      "step": 14542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.657409978709618e-05,
+      "loss": 0.0167,
+      "step": 14544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6569471443117655e-05,
+      "loss": 0.0109,
+      "step": 14546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.656484309913913e-05,
+      "loss": 0.7003,
+      "step": 14548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6560214755160606e-05,
+      "loss": 0.3928,
+      "step": 14550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.655558641118208e-05,
+      "loss": 0.002,
+      "step": 14552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6550958067203556e-05,
+      "loss": 2.2274,
+      "step": 14554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6546329723225032e-05,
+      "loss": 6.2223,
+      "step": 14556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6541701379246507e-05,
+      "loss": 1.9749,
+      "step": 14558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6537073035267983e-05,
+      "loss": 5.9238,
+      "step": 14560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6532444691289458e-05,
+      "loss": 0.3642,
+      "step": 14562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6527816347310934e-05,
+      "loss": 0.0018,
+      "step": 14564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6523188003332406e-05,
+      "loss": 0.8708,
+      "step": 14566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.651855965935388e-05,
+      "loss": 2.196,
+      "step": 14568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.651393131537536e-05,
+      "loss": 4.1129,
+      "step": 14570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6509302971396835e-05,
+      "loss": 2.0629,
+      "step": 14572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.650467462741831e-05,
+      "loss": 2.4866,
+      "step": 14574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6500046283439786e-05,
+      "loss": 0.731,
+      "step": 14576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.649541793946126e-05,
+      "loss": 0.0461,
+      "step": 14578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6490789595482737e-05,
+      "loss": 0.8755,
+      "step": 14580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6486161251504212e-05,
+      "loss": 0.8764,
+      "step": 14582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6481532907525688e-05,
+      "loss": 2.6596,
+      "step": 14584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6476904563547163e-05,
+      "loss": 2.0307,
+      "step": 14586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.647227621956864e-05,
+      "loss": 4.654,
+      "step": 14588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6467647875590114e-05,
+      "loss": 2.2553,
+      "step": 14590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.646301953161159e-05,
+      "loss": 5.778,
+      "step": 14592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6458391187633065e-05,
+      "loss": 5.6582,
+      "step": 14594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.645376284365454e-05,
+      "loss": 0.7101,
+      "step": 14596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6449134499676016e-05,
+      "loss": 6.4878,
+      "step": 14598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.644450615569749e-05,
+      "loss": 1.7383,
+      "step": 14600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6439877811718967e-05,
+      "loss": 2.8553,
+      "step": 14602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6435249467740442e-05,
+      "loss": 0.0021,
+      "step": 14604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6430621123761918e-05,
+      "loss": 1.7971,
+      "step": 14606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6425992779783393e-05,
+      "loss": 0.4776,
+      "step": 14608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6421364435804872e-05,
+      "loss": 1.4232,
+      "step": 14610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6416736091826347e-05,
+      "loss": 2.9551,
+      "step": 14612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6412107747847823e-05,
+      "loss": 4.4844,
+      "step": 14614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6407479403869298e-05,
+      "loss": 0.3095,
+      "step": 14616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6402851059890773e-05,
+      "loss": 0.5737,
+      "step": 14618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.639822271591225e-05,
+      "loss": 0.4895,
+      "step": 14620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6393594371933724e-05,
+      "loss": 6.6313,
+      "step": 14622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.63889660279552e-05,
+      "loss": 0.8642,
+      "step": 14624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6384337683976675e-05,
+      "loss": 3.9628,
+      "step": 14626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.637970933999815e-05,
+      "loss": 0.1049,
+      "step": 14628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6375080996019626e-05,
+      "loss": 5.5035,
+      "step": 14630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.63704526520411e-05,
+      "loss": 2.4347,
+      "step": 14632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6365824308062577e-05,
+      "loss": 1.3342,
+      "step": 14634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6361195964084052e-05,
+      "loss": 5.1133,
+      "step": 14636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6356567620105524e-05,
+      "loss": 4.7734,
+      "step": 14638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6351939276127e-05,
+      "loss": 1.2099,
+      "step": 14640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.634731093214848e-05,
+      "loss": 1.3096,
+      "step": 14642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6342682588169954e-05,
+      "loss": 0.174,
+      "step": 14644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.633805424419143e-05,
+      "loss": 5.1999,
+      "step": 14646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6333425900212905e-05,
+      "loss": 0.3116,
+      "step": 14648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.632879755623438e-05,
+      "loss": 2.0577,
+      "step": 14650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6324169212255856e-05,
+      "loss": 0.1741,
+      "step": 14652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.631954086827733e-05,
+      "loss": 0.0057,
+      "step": 14654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6314912524298807e-05,
+      "loss": 1.2336,
+      "step": 14656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6310284180320282e-05,
+      "loss": 0.7137,
+      "step": 14658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6305655836341757e-05,
+      "loss": 3.854,
+      "step": 14660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6301027492363233e-05,
+      "loss": 0.2649,
+      "step": 14662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6296399148384708e-05,
+      "loss": 1.3485,
+      "step": 14664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6291770804406184e-05,
+      "loss": 0.7348,
+      "step": 14666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.628714246042766e-05,
+      "loss": 2.0194,
+      "step": 14668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6282514116449134e-05,
+      "loss": 0.7178,
+      "step": 14670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.627788577247061e-05,
+      "loss": 1.2336,
+      "step": 14672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6273257428492085e-05,
+      "loss": 1.1309,
+      "step": 14674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.626862908451356e-05,
+      "loss": 0.0027,
+      "step": 14676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6264000740535036e-05,
+      "loss": 1.1177,
+      "step": 14678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.625937239655651e-05,
+      "loss": 3.6393,
+      "step": 14680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6254744052577987e-05,
+      "loss": 0.9661,
+      "step": 14682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6250115708599466e-05,
+      "loss": 2.6503,
+      "step": 14684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.624548736462094e-05,
+      "loss": 4.5529,
+      "step": 14686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6240859020642417e-05,
+      "loss": 0.5666,
+      "step": 14688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6236230676663892e-05,
+      "loss": 0.1965,
+      "step": 14690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6231602332685368e-05,
+      "loss": 3.2071,
+      "step": 14692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6226973988706843e-05,
+      "loss": 0.9101,
+      "step": 14694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.622234564472832e-05,
+      "loss": 1.8072,
+      "step": 14696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6217717300749794e-05,
+      "loss": 0.0004,
+      "step": 14698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.621308895677127e-05,
+      "loss": 1.6776,
+      "step": 14700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6208460612792745e-05,
+      "loss": 1.3997,
+      "step": 14702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.620383226881422e-05,
+      "loss": 1.2409,
+      "step": 14704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6199203924835696e-05,
+      "loss": 0.0426,
+      "step": 14706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.619457558085717e-05,
+      "loss": 2.6735,
+      "step": 14708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6189947236878643e-05,
+      "loss": 2.1446,
+      "step": 14710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.618531889290012e-05,
+      "loss": 1.559,
+      "step": 14712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6180690548921594e-05,
+      "loss": 1.3772,
+      "step": 14714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6176062204943073e-05,
+      "loss": 0.3061,
+      "step": 14716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6171433860964548e-05,
+      "loss": 6.0269,
+      "step": 14718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6166805516986023e-05,
+      "loss": 6.1993,
+      "step": 14720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.61621771730075e-05,
+      "loss": 2.9988,
+      "step": 14722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6157548829028974e-05,
+      "loss": 0.8186,
+      "step": 14724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.615292048505045e-05,
+      "loss": 4.1417,
+      "step": 14726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6148292141071925e-05,
+      "loss": 0.9359,
+      "step": 14728
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.61436637970934e-05,
+      "loss": 2.2458,
+      "step": 14730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6139035453114876e-05,
+      "loss": 0.0007,
+      "step": 14732
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.613440710913635e-05,
+      "loss": 3.3542,
+      "step": 14734
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6129778765157827e-05,
+      "loss": 4.3072,
+      "step": 14736
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6125150421179302e-05,
+      "loss": 0.7174,
+      "step": 14738
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6120522077200778e-05,
+      "loss": 2.8969,
+      "step": 14740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6115893733222253e-05,
+      "loss": 0.002,
+      "step": 14742
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.611126538924373e-05,
+      "loss": 1.7346,
+      "step": 14744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6106637045265204e-05,
+      "loss": 1.8579,
+      "step": 14746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.610200870128668e-05,
+      "loss": 2.2357,
+      "step": 14748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6097380357308155e-05,
+      "loss": 1.0897,
+      "step": 14750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.609275201332963e-05,
+      "loss": 0.3098,
+      "step": 14752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6088123669351106e-05,
+      "loss": 2.2993,
+      "step": 14754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6083495325372585e-05,
+      "loss": 1.9363,
+      "step": 14756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.607886698139406e-05,
+      "loss": 2.1081,
+      "step": 14758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6074238637415535e-05,
+      "loss": 1.4716,
+      "step": 14760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.606961029343701e-05,
+      "loss": 0.0005,
+      "step": 14762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6064981949458486e-05,
+      "loss": 1.3264,
+      "step": 14764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.606035360547996e-05,
+      "loss": 0.2899,
+      "step": 14766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6055725261501437e-05,
+      "loss": 4.7377,
+      "step": 14768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6051096917522912e-05,
+      "loss": 5.0884,
+      "step": 14770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6046468573544388e-05,
+      "loss": 4.0877,
+      "step": 14772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6041840229565863e-05,
+      "loss": 2.2017,
+      "step": 14774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.603721188558734e-05,
+      "loss": 0.3592,
+      "step": 14776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6032583541608814e-05,
+      "loss": 1.7904,
+      "step": 14778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.602795519763029e-05,
+      "loss": 5.972,
+      "step": 14780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.602332685365176e-05,
+      "loss": 0.1436,
+      "step": 14782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6018698509673237e-05,
+      "loss": 0.5674,
+      "step": 14784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6014070165694712e-05,
+      "loss": 0.3908,
+      "step": 14786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6009441821716188e-05,
+      "loss": 1.2218,
+      "step": 14788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6004813477737667e-05,
+      "loss": 0.799,
+      "step": 14790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6000185133759142e-05,
+      "loss": 0.5871,
+      "step": 14792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5995556789780618e-05,
+      "loss": 6.7282,
+      "step": 14794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5990928445802093e-05,
+      "loss": 0.6341,
+      "step": 14796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.598630010182357e-05,
+      "loss": 0.29,
+      "step": 14798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5981671757845044e-05,
+      "loss": 5.2718,
+      "step": 14800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.597704341386652e-05,
+      "loss": 2.4817,
+      "step": 14802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5972415069887995e-05,
+      "loss": 0.0114,
+      "step": 14804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.596778672590947e-05,
+      "loss": 1.8872,
+      "step": 14806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5963158381930946e-05,
+      "loss": 2.8608,
+      "step": 14808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.595853003795242e-05,
+      "loss": 0.9467,
+      "step": 14810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5953901693973896e-05,
+      "loss": 3.1035,
+      "step": 14812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5949273349995372e-05,
+      "loss": 0.6448,
+      "step": 14814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5944645006016847e-05,
+      "loss": 0.0015,
+      "step": 14816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5940016662038323e-05,
+      "loss": 4.2871,
+      "step": 14818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5935388318059798e-05,
+      "loss": 7.0314,
+      "step": 14820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5930759974081274e-05,
+      "loss": 0.0866,
+      "step": 14822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.592613163010275e-05,
+      "loss": 1.7823,
+      "step": 14824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5921503286124224e-05,
+      "loss": 0.2491,
+      "step": 14826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.59168749421457e-05,
+      "loss": 1.9227,
+      "step": 14828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.591224659816718e-05,
+      "loss": 1.9496,
+      "step": 14830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5907618254188654e-05,
+      "loss": 1.9059,
+      "step": 14832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.590298991021013e-05,
+      "loss": 2.865,
+      "step": 14834
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5898361566231605e-05,
+      "loss": 1.7889,
+      "step": 14836
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.589373322225308e-05,
+      "loss": 0.8653,
+      "step": 14838
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5889104878274556e-05,
+      "loss": 0.0018,
+      "step": 14840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.588447653429603e-05,
+      "loss": 3.8934,
+      "step": 14842
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5879848190317507e-05,
+      "loss": 2.5471,
+      "step": 14844
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5875219846338982e-05,
+      "loss": 0.9303,
+      "step": 14846
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5870591502360457e-05,
+      "loss": 1.5211,
+      "step": 14848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5865963158381933e-05,
+      "loss": 1.9147,
+      "step": 14850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5861334814403408e-05,
+      "loss": 0.417,
+      "step": 14852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5856706470424884e-05,
+      "loss": 1.183,
+      "step": 14854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5852078126446356e-05,
+      "loss": 1.8229,
+      "step": 14856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.584744978246783e-05,
+      "loss": 1.2163,
+      "step": 14858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5842821438489307e-05,
+      "loss": 0.5313,
+      "step": 14860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5838193094510782e-05,
+      "loss": 7.5661,
+      "step": 14862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.583356475053226e-05,
+      "loss": 0.3784,
+      "step": 14864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5828936406553736e-05,
+      "loss": 1.8726,
+      "step": 14866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.582430806257521e-05,
+      "loss": 0.0016,
+      "step": 14868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5819679718596687e-05,
+      "loss": 2.6977,
+      "step": 14870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5815051374618163e-05,
+      "loss": 0.9002,
+      "step": 14872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5810423030639638e-05,
+      "loss": 0.6695,
+      "step": 14874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5805794686661113e-05,
+      "loss": 2.54,
+      "step": 14876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.580116634268259e-05,
+      "loss": 4.1447,
+      "step": 14878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5796537998704064e-05,
+      "loss": 0.0044,
+      "step": 14880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.579190965472554e-05,
+      "loss": 0.8857,
+      "step": 14882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5787281310747015e-05,
+      "loss": 1.6281,
+      "step": 14884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.578265296676849e-05,
+      "loss": 4.4905,
+      "step": 14886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5778024622789966e-05,
+      "loss": 1.1369,
+      "step": 14888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.577339627881144e-05,
+      "loss": 0.0006,
+      "step": 14890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5768767934832917e-05,
+      "loss": 4.5379,
+      "step": 14892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5764139590854392e-05,
+      "loss": 0.4364,
+      "step": 14894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5759511246875868e-05,
+      "loss": 0.0557,
+      "step": 14896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5754882902897343e-05,
+      "loss": 1.1453,
+      "step": 14898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.575025455891882e-05,
+      "loss": 2.5199,
+      "step": 14900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5745626214940294e-05,
+      "loss": 3.7374,
+      "step": 14902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5740997870961773e-05,
+      "loss": 4.4743,
+      "step": 14904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5736369526983248e-05,
+      "loss": 0.0123,
+      "step": 14906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5731741183004724e-05,
+      "loss": 0.167,
+      "step": 14908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.57271128390262e-05,
+      "loss": 2.6547,
+      "step": 14910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5722484495047674e-05,
+      "loss": 1.7532,
+      "step": 14912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.571785615106915e-05,
+      "loss": 3.5342,
+      "step": 14914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5713227807090625e-05,
+      "loss": 4.8889,
+      "step": 14916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.57085994631121e-05,
+      "loss": 3.4013,
+      "step": 14918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5703971119133576e-05,
+      "loss": 2.7721,
+      "step": 14920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.569934277515505e-05,
+      "loss": 1.8611,
+      "step": 14922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5694714431176527e-05,
+      "loss": 0.0009,
+      "step": 14924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5690086087198002e-05,
+      "loss": 1.7731,
+      "step": 14926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5685457743219474e-05,
+      "loss": 0.7718,
+      "step": 14928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.568082939924095e-05,
+      "loss": 1.0689,
+      "step": 14930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5676201055262425e-05,
+      "loss": 2.8077,
+      "step": 14932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.56715727112839e-05,
+      "loss": 2.9833,
+      "step": 14934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.566694436730538e-05,
+      "loss": 0.9209,
+      "step": 14936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5662316023326855e-05,
+      "loss": 2.9415,
+      "step": 14938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.565768767934833e-05,
+      "loss": 2.8763,
+      "step": 14940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5653059335369806e-05,
+      "loss": 6.0491,
+      "step": 14942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.564843099139128e-05,
+      "loss": 1.1783,
+      "step": 14944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5643802647412757e-05,
+      "loss": 1.5141,
+      "step": 14946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5639174303434232e-05,
+      "loss": 0.447,
+      "step": 14948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5634545959455707e-05,
+      "loss": 0.584,
+      "step": 14950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5629917615477183e-05,
+      "loss": 1.4335,
+      "step": 14952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.562528927149866e-05,
+      "loss": 1.4051,
+      "step": 14954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5620660927520134e-05,
+      "loss": 1.2567,
+      "step": 14956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.561603258354161e-05,
+      "loss": 2.5531,
+      "step": 14958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5611404239563085e-05,
+      "loss": 1.7414,
+      "step": 14960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.560677589558456e-05,
+      "loss": 0.8479,
+      "step": 14962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5602147551606035e-05,
+      "loss": 0.6427,
+      "step": 14964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.559751920762751e-05,
+      "loss": 1.2743,
+      "step": 14966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5592890863648986e-05,
+      "loss": 3.4845,
+      "step": 14968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5588262519670462e-05,
+      "loss": 1.5812,
+      "step": 14970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5583634175691937e-05,
+      "loss": 0.9783,
+      "step": 14972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5579005831713413e-05,
+      "loss": 1.109,
+      "step": 14974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.557437748773489e-05,
+      "loss": 2.2722,
+      "step": 14976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5569749143756367e-05,
+      "loss": 0.8635,
+      "step": 14978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5565120799777842e-05,
+      "loss": 2.9446,
+      "step": 14980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5560492455799318e-05,
+      "loss": 1.0367,
+      "step": 14982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5555864111820793e-05,
+      "loss": 0.2211,
+      "step": 14984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.555123576784227e-05,
+      "loss": 2.8136,
+      "step": 14986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5546607423863744e-05,
+      "loss": 2.019,
+      "step": 14988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.554197907988522e-05,
+      "loss": 1.6492,
+      "step": 14990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5537350735906695e-05,
+      "loss": 1.0376,
+      "step": 14992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.553272239192817e-05,
+      "loss": 0.0006,
+      "step": 14994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5528094047949646e-05,
+      "loss": 0.4591,
+      "step": 14996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.552346570397112e-05,
+      "loss": 0.0004,
+      "step": 14998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5518837359992593e-05,
+      "loss": 4.7593,
+      "step": 15000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.551420901601407e-05,
+      "loss": 4.4031,
+      "step": 15002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5509580672035544e-05,
+      "loss": 2.078,
+      "step": 15004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.550495232805702e-05,
+      "loss": 0.8005,
+      "step": 15006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5500323984078495e-05,
+      "loss": 3.2617,
+      "step": 15008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5495695640099974e-05,
+      "loss": 0.0712,
+      "step": 15010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.549106729612145e-05,
+      "loss": 8.2118,
+      "step": 15012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5486438952142924e-05,
+      "loss": 2.207,
+      "step": 15014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.54818106081644e-05,
+      "loss": 0.0015,
+      "step": 15016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5477182264185875e-05,
+      "loss": 2.6837,
+      "step": 15018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.547255392020735e-05,
+      "loss": 0.8953,
+      "step": 15020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5467925576228826e-05,
+      "loss": 2.1521,
+      "step": 15022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.54632972322503e-05,
+      "loss": 1.2342,
+      "step": 15024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5458668888271777e-05,
+      "loss": 0.4517,
+      "step": 15026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5454040544293252e-05,
+      "loss": 2.7703,
+      "step": 15028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5449412200314728e-05,
+      "loss": 1.1862,
+      "step": 15030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5444783856336203e-05,
+      "loss": 0.3233,
+      "step": 15032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.544015551235768e-05,
+      "loss": 0.0008,
+      "step": 15034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5435527168379154e-05,
+      "loss": 3.1864,
+      "step": 15036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.543089882440063e-05,
+      "loss": 3.3178,
+      "step": 15038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5426270480422105e-05,
+      "loss": 0.4009,
+      "step": 15040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.542164213644358e-05,
+      "loss": 0.3432,
+      "step": 15042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5417013792465056e-05,
+      "loss": 6.693,
+      "step": 15044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.541238544848653e-05,
+      "loss": 3.753,
+      "step": 15046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5407757104508007e-05,
+      "loss": 0.4408,
+      "step": 15048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5403128760529485e-05,
+      "loss": 0.5058,
+      "step": 15050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.539850041655096e-05,
+      "loss": 0.0007,
+      "step": 15052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5393872072572436e-05,
+      "loss": 0.9045,
+      "step": 15054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5389243728593912e-05,
+      "loss": 1.6803,
+      "step": 15056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 3.0789,
+      "step": 15058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5379987040636863e-05,
+      "loss": 2.062,
+      "step": 15060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5375358696658338e-05,
+      "loss": 2.3429,
+      "step": 15062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5370730352679813e-05,
+      "loss": 0.0027,
+      "step": 15064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.536610200870129e-05,
+      "loss": 2.4577,
+      "step": 15066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5361473664722764e-05,
+      "loss": 0.4035,
+      "step": 15068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.535684532074424e-05,
+      "loss": 5.5753,
+      "step": 15070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5352216976765712e-05,
+      "loss": 0.6447,
+      "step": 15072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5347588632787187e-05,
+      "loss": 1.3972,
+      "step": 15074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5342960288808663e-05,
+      "loss": 0.7886,
+      "step": 15076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5338331944830138e-05,
+      "loss": 0.0246,
+      "step": 15078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5333703600851613e-05,
+      "loss": 0.1108,
+      "step": 15080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.532907525687309e-05,
+      "loss": 0.8593,
+      "step": 15082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5324446912894568e-05,
+      "loss": 0.9502,
+      "step": 15084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5319818568916043e-05,
+      "loss": 1.7536,
+      "step": 15086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.531519022493752e-05,
+      "loss": 7.9619,
+      "step": 15088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5310561880958994e-05,
+      "loss": 0.3611,
+      "step": 15090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.530593353698047e-05,
+      "loss": 5.1475,
+      "step": 15092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5301305193001945e-05,
+      "loss": 2.215,
+      "step": 15094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.529667684902342e-05,
+      "loss": 0.0014,
+      "step": 15096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5292048505044896e-05,
+      "loss": 0.4632,
+      "step": 15098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.528742016106637e-05,
+      "loss": 2.9346,
+      "step": 15100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5282791817087847e-05,
+      "loss": 1.3603,
+      "step": 15102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5278163473109322e-05,
+      "loss": 0.0003,
+      "step": 15104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5273535129130797e-05,
+      "loss": 2.4566,
+      "step": 15106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5268906785152273e-05,
+      "loss": 1.2595,
+      "step": 15108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5264278441173748e-05,
+      "loss": 2.9438,
+      "step": 15110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5259650097195224e-05,
+      "loss": 0.0203,
+      "step": 15112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.52550217532167e-05,
+      "loss": 0.2005,
+      "step": 15114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5250393409238176e-05,
+      "loss": 1.3519,
+      "step": 15116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5245765065259652e-05,
+      "loss": 0.3705,
+      "step": 15118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5241136721281127e-05,
+      "loss": 3.411,
+      "step": 15120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5236508377302602e-05,
+      "loss": 4.527,
+      "step": 15122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5231880033324078e-05,
+      "loss": 1.4769,
+      "step": 15124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5227251689345553e-05,
+      "loss": 0.6804,
+      "step": 15126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5222623345367029e-05,
+      "loss": 0.2117,
+      "step": 15128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5217995001388504e-05,
+      "loss": 1.1775,
+      "step": 15130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.521336665740998e-05,
+      "loss": 0.8773,
+      "step": 15132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5208738313431457e-05,
+      "loss": 1.8963,
+      "step": 15134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5204109969452932e-05,
+      "loss": 2.9274,
+      "step": 15136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5199481625474408e-05,
+      "loss": 0.5219,
+      "step": 15138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5194853281495883e-05,
+      "loss": 0.8237,
+      "step": 15140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5190224937517358e-05,
+      "loss": 1.0221,
+      "step": 15142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5185596593538834e-05,
+      "loss": 1.384,
+      "step": 15144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5180968249560308e-05,
+      "loss": 0.1141,
+      "step": 15146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5176339905581783e-05,
+      "loss": 2.7436,
+      "step": 15148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5171711561603258e-05,
+      "loss": 0.0956,
+      "step": 15150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5167083217624734e-05,
+      "loss": 0.7073,
+      "step": 15152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.516245487364621e-05,
+      "loss": 0.0009,
+      "step": 15154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5157826529667685e-05,
+      "loss": 4.3875,
+      "step": 15156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.515319818568916e-05,
+      "loss": 5.8297,
+      "step": 15158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5148569841710636e-05,
+      "loss": 0.2733,
+      "step": 15160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5143941497732111e-05,
+      "loss": 1.9004,
+      "step": 15162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5139313153753586e-05,
+      "loss": 0.0005,
+      "step": 15164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5134684809775063e-05,
+      "loss": 0.6707,
+      "step": 15166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5130056465796539e-05,
+      "loss": 0.8294,
+      "step": 15168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5125428121818014e-05,
+      "loss": 2.0475,
+      "step": 15170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.512079977783949e-05,
+      "loss": 1.2308,
+      "step": 15172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5116171433860965e-05,
+      "loss": 0.5062,
+      "step": 15174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.511154308988244e-05,
+      "loss": 6.1508,
+      "step": 15176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5106914745903916e-05,
+      "loss": 4.5625,
+      "step": 15178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5102286401925391e-05,
+      "loss": 0.1493,
+      "step": 15180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5097658057946867e-05,
+      "loss": 0.979,
+      "step": 15182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5093029713968342e-05,
+      "loss": 0.6347,
+      "step": 15184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5088401369989818e-05,
+      "loss": 0.6137,
+      "step": 15186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5083773026011295e-05,
+      "loss": 0.0011,
+      "step": 15188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.507914468203277e-05,
+      "loss": 1.8091,
+      "step": 15190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5074516338054246e-05,
+      "loss": 1.7792,
+      "step": 15192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5069887994075721e-05,
+      "loss": 5.2022,
+      "step": 15194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5065259650097197e-05,
+      "loss": 0.2749,
+      "step": 15196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5060631306118672e-05,
+      "loss": 0.7508,
+      "step": 15198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5056002962140147e-05,
+      "loss": 4.1284,
+      "step": 15200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5051374618161623e-05,
+      "loss": 0.0007,
+      "step": 15202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5046746274183098e-05,
+      "loss": 1.8323,
+      "step": 15204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5042117930204574e-05,
+      "loss": 0.0008,
+      "step": 15206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.503748958622605e-05,
+      "loss": 0.9159,
+      "step": 15208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5032861242247526e-05,
+      "loss": 0.3176,
+      "step": 15210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5028232898269002e-05,
+      "loss": 0.1621,
+      "step": 15212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5023604554290477e-05,
+      "loss": 0.3267,
+      "step": 15214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5018976210311952e-05,
+      "loss": 3.8394,
+      "step": 15216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5014347866333425e-05,
+      "loss": 0.0036,
+      "step": 15218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5009719522354902e-05,
+      "loss": 1.6717,
+      "step": 15220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5005091178376377e-05,
+      "loss": 3.3102,
+      "step": 15222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.5000462834397852e-05,
+      "loss": 0.0023,
+      "step": 15224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4995834490419328e-05,
+      "loss": 0.1041,
+      "step": 15226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4991206146440803e-05,
+      "loss": 0.4419,
+      "step": 15228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4986577802462279e-05,
+      "loss": 0.0018,
+      "step": 15230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4981949458483754e-05,
+      "loss": 7.1431,
+      "step": 15232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.497732111450523e-05,
+      "loss": 1.955,
+      "step": 15234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4972692770526705e-05,
+      "loss": 1.8349,
+      "step": 15236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.496806442654818e-05,
+      "loss": 0.0261,
+      "step": 15238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4963436082569658e-05,
+      "loss": 0.2695,
+      "step": 15240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4958807738591133e-05,
+      "loss": 0.3752,
+      "step": 15242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4954179394612608e-05,
+      "loss": 3.3142,
+      "step": 15244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4949551050634084e-05,
+      "loss": 0.9772,
+      "step": 15246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.494492270665556e-05,
+      "loss": 1.4415,
+      "step": 15248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4940294362677035e-05,
+      "loss": 0.9614,
+      "step": 15250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.493566601869851e-05,
+      "loss": 3.4091,
+      "step": 15252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4931037674719986e-05,
+      "loss": 0.0034,
+      "step": 15254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4926409330741461e-05,
+      "loss": 0.0015,
+      "step": 15256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4921780986762936e-05,
+      "loss": 1.1872,
+      "step": 15258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4917152642784414e-05,
+      "loss": 0.2279,
+      "step": 15260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4912524298805889e-05,
+      "loss": 0.0008,
+      "step": 15262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4907895954827364e-05,
+      "loss": 1.0994,
+      "step": 15264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.490326761084884e-05,
+      "loss": 3.6229,
+      "step": 15266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4898639266870315e-05,
+      "loss": 1.8237,
+      "step": 15268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.489401092289179e-05,
+      "loss": 2.1466,
+      "step": 15270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4889382578913266e-05,
+      "loss": 0.0007,
+      "step": 15272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4884754234934742e-05,
+      "loss": 3.9632,
+      "step": 15274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4880125890956217e-05,
+      "loss": 0.0834,
+      "step": 15276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4875497546977692e-05,
+      "loss": 4.6524,
+      "step": 15278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.487086920299917e-05,
+      "loss": 0.0007,
+      "step": 15280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4866240859020645e-05,
+      "loss": 4.8768,
+      "step": 15282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.486161251504212e-05,
+      "loss": 4.5969,
+      "step": 15284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4856984171063596e-05,
+      "loss": 0.3139,
+      "step": 15286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4852355827085071e-05,
+      "loss": 0.0004,
+      "step": 15288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4847727483106543e-05,
+      "loss": 0.1105,
+      "step": 15290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4843099139128019e-05,
+      "loss": 1.3638,
+      "step": 15292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4838470795149496e-05,
+      "loss": 3.459,
+      "step": 15294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4833842451170971e-05,
+      "loss": 1.9777,
+      "step": 15296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4829214107192447e-05,
+      "loss": 1.6767,
+      "step": 15298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4824585763213922e-05,
+      "loss": 2.4635,
+      "step": 15300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4819957419235397e-05,
+      "loss": 2.2707,
+      "step": 15302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4815329075256873e-05,
+      "loss": 0.1486,
+      "step": 15304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4810700731278348e-05,
+      "loss": 1.4079,
+      "step": 15306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4806072387299824e-05,
+      "loss": 2.5098,
+      "step": 15308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4801444043321299e-05,
+      "loss": 0.1462,
+      "step": 15310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4796815699342775e-05,
+      "loss": 0.5792,
+      "step": 15312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4792187355364252e-05,
+      "loss": 2.6547,
+      "step": 15314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4787559011385727e-05,
+      "loss": 4.0119,
+      "step": 15316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4782930667407203e-05,
+      "loss": 1.0354,
+      "step": 15318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4778302323428678e-05,
+      "loss": 1.8484,
+      "step": 15320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4773673979450153e-05,
+      "loss": 2.1038,
+      "step": 15322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4769045635471629e-05,
+      "loss": 0.2219,
+      "step": 15324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4764417291493104e-05,
+      "loss": 1.8255,
+      "step": 15326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.475978894751458e-05,
+      "loss": 3.1899,
+      "step": 15328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4755160603536055e-05,
+      "loss": 3.6035,
+      "step": 15330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.475053225955753e-05,
+      "loss": 1.4895,
+      "step": 15332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4745903915579008e-05,
+      "loss": 4.164,
+      "step": 15334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4741275571600483e-05,
+      "loss": 1.4289,
+      "step": 15336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4736647227621958e-05,
+      "loss": 4.2977,
+      "step": 15338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4732018883643434e-05,
+      "loss": 4.9745,
+      "step": 15340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.472739053966491e-05,
+      "loss": 1.4664,
+      "step": 15342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4722762195686385e-05,
+      "loss": 6.3983,
+      "step": 15344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.471813385170786e-05,
+      "loss": 0.1655,
+      "step": 15346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4713505507729336e-05,
+      "loss": 0.8937,
+      "step": 15348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4708877163750811e-05,
+      "loss": 0.026,
+      "step": 15350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4704248819772286e-05,
+      "loss": 0.001,
+      "step": 15352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4699620475793764e-05,
+      "loss": 0.5701,
+      "step": 15354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4694992131815239e-05,
+      "loss": 1.8426,
+      "step": 15356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4690363787836714e-05,
+      "loss": 4.6585,
+      "step": 15358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.468573544385819e-05,
+      "loss": 3.1048,
+      "step": 15360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4681107099879662e-05,
+      "loss": 2.3582,
+      "step": 15362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4676478755901137e-05,
+      "loss": 0.0032,
+      "step": 15364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4671850411922613e-05,
+      "loss": 1.4687,
+      "step": 15366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.466722206794409e-05,
+      "loss": 0.001,
+      "step": 15368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4662593723965565e-05,
+      "loss": 4.0862,
+      "step": 15370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.465796537998704e-05,
+      "loss": 0.2431,
+      "step": 15372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4653337036008516e-05,
+      "loss": 0.0169,
+      "step": 15374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4648708692029992e-05,
+      "loss": 3.369,
+      "step": 15376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4644080348051467e-05,
+      "loss": 2.2996,
+      "step": 15378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4639452004072942e-05,
+      "loss": 0.5851,
+      "step": 15380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4634823660094418e-05,
+      "loss": 1.1961,
+      "step": 15382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4630195316115893e-05,
+      "loss": 3.1466,
+      "step": 15384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4625566972137369e-05,
+      "loss": 1.0878,
+      "step": 15386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4620938628158846e-05,
+      "loss": 2.5726,
+      "step": 15388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4616310284180321e-05,
+      "loss": 1.3574,
+      "step": 15390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4611681940201797e-05,
+      "loss": 0.9666,
+      "step": 15392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4607053596223272e-05,
+      "loss": 0.5896,
+      "step": 15394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4602425252244747e-05,
+      "loss": 0.0021,
+      "step": 15396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4597796908266223e-05,
+      "loss": 1.5038,
+      "step": 15398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4593168564287698e-05,
+      "loss": 0.0622,
+      "step": 15400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4588540220309174e-05,
+      "loss": 0.873,
+      "step": 15402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.458391187633065e-05,
+      "loss": 2.4386,
+      "step": 15404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4579283532352125e-05,
+      "loss": 5.1184,
+      "step": 15406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4574655188373602e-05,
+      "loss": 7.5952,
+      "step": 15408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4570026844395077e-05,
+      "loss": 0.007,
+      "step": 15410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4565398500416553e-05,
+      "loss": 0.9466,
+      "step": 15412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4560770156438028e-05,
+      "loss": 3.1908,
+      "step": 15414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4556141812459503e-05,
+      "loss": 7.3198,
+      "step": 15416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4551513468480979e-05,
+      "loss": 2.0221,
+      "step": 15418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4546885124502454e-05,
+      "loss": 0.0004,
+      "step": 15420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.454225678052393e-05,
+      "loss": 0.0018,
+      "step": 15422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4537628436545405e-05,
+      "loss": 0.0014,
+      "step": 15424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.453300009256688e-05,
+      "loss": 0.7562,
+      "step": 15426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4528371748588358e-05,
+      "loss": 0.0006,
+      "step": 15428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4523743404609833e-05,
+      "loss": 0.4712,
+      "step": 15430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4519115060631309e-05,
+      "loss": 0.0033,
+      "step": 15432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4514486716652784e-05,
+      "loss": 0.1669,
+      "step": 15434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4509858372674256e-05,
+      "loss": 2.1042,
+      "step": 15436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4505230028695731e-05,
+      "loss": 1.7426,
+      "step": 15438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4500601684717209e-05,
+      "loss": 1.3541,
+      "step": 15440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4495973340738684e-05,
+      "loss": 0.7745,
+      "step": 15442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.449134499676016e-05,
+      "loss": 0.6909,
+      "step": 15444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4486716652781635e-05,
+      "loss": 0.4365,
+      "step": 15446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.448208830880311e-05,
+      "loss": 2.4793,
+      "step": 15448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4477459964824586e-05,
+      "loss": 1.0311,
+      "step": 15450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4472831620846061e-05,
+      "loss": 0.0002,
+      "step": 15452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4468203276867536e-05,
+      "loss": 0.0003,
+      "step": 15454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4463574932889012e-05,
+      "loss": 6.2751,
+      "step": 15456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4458946588910487e-05,
+      "loss": 5.8581,
+      "step": 15458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4454318244931964e-05,
+      "loss": 3.5022,
+      "step": 15460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.444968990095344e-05,
+      "loss": 0.0004,
+      "step": 15462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4445061556974915e-05,
+      "loss": 0.0003,
+      "step": 15464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.444043321299639e-05,
+      "loss": 1.3267,
+      "step": 15466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4435804869017866e-05,
+      "loss": 0.7139,
+      "step": 15468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4431176525039342e-05,
+      "loss": 1.3044,
+      "step": 15470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4426548181060817e-05,
+      "loss": 2.6736,
+      "step": 15472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4421919837082292e-05,
+      "loss": 5.14,
+      "step": 15474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4417291493103768e-05,
+      "loss": 0.2602,
+      "step": 15476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4412663149125243e-05,
+      "loss": 0.2546,
+      "step": 15478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.440803480514672e-05,
+      "loss": 1.2572,
+      "step": 15480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4403406461168196e-05,
+      "loss": 2.4543,
+      "step": 15482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4398778117189671e-05,
+      "loss": 0.629,
+      "step": 15484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4394149773211147e-05,
+      "loss": 4.9821,
+      "step": 15486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4389521429232622e-05,
+      "loss": 1.0008,
+      "step": 15488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4384893085254098e-05,
+      "loss": 1.0793,
+      "step": 15490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4380264741275573e-05,
+      "loss": 0.0829,
+      "step": 15492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4375636397297048e-05,
+      "loss": 4.233,
+      "step": 15494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4371008053318524e-05,
+      "loss": 0.0005,
+      "step": 15496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.436637970934e-05,
+      "loss": 2.071,
+      "step": 15498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4361751365361475e-05,
+      "loss": 0.042,
+      "step": 15500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4357123021382952e-05,
+      "loss": 2.921,
+      "step": 15502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4352494677404427e-05,
+      "loss": 2.0097,
+      "step": 15504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4347866333425903e-05,
+      "loss": 5.3953,
+      "step": 15506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4343237989447375e-05,
+      "loss": 1.2576,
+      "step": 15508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.433860964546885e-05,
+      "loss": 2.8586,
+      "step": 15510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4333981301490325e-05,
+      "loss": 2.0089,
+      "step": 15512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4329352957511803e-05,
+      "loss": 0.3712,
+      "step": 15514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4324724613533278e-05,
+      "loss": 2.9465,
+      "step": 15516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4320096269554753e-05,
+      "loss": 1.4289,
+      "step": 15518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4315467925576229e-05,
+      "loss": 0.4502,
+      "step": 15520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4310839581597704e-05,
+      "loss": 5.2281,
+      "step": 15522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.430621123761918e-05,
+      "loss": 1.0918,
+      "step": 15524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4301582893640655e-05,
+      "loss": 0.9605,
+      "step": 15526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.429695454966213e-05,
+      "loss": 2.6894,
+      "step": 15528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4292326205683606e-05,
+      "loss": 3.6249,
+      "step": 15530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4287697861705081e-05,
+      "loss": 1.8281,
+      "step": 15532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4283069517726559e-05,
+      "loss": 2.181,
+      "step": 15534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4278441173748034e-05,
+      "loss": 0.8032,
+      "step": 15536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.427381282976951e-05,
+      "loss": 1.1125,
+      "step": 15538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4269184485790985e-05,
+      "loss": 1.8308,
+      "step": 15540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.426455614181246e-05,
+      "loss": 2.2997,
+      "step": 15542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4259927797833936e-05,
+      "loss": 0.9325,
+      "step": 15544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4255299453855411e-05,
+      "loss": 1.4333,
+      "step": 15546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4250671109876887e-05,
+      "loss": 2.7233,
+      "step": 15548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4246042765898362e-05,
+      "loss": 0.5034,
+      "step": 15550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4241414421919837e-05,
+      "loss": 2.8277,
+      "step": 15552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4236786077941314e-05,
+      "loss": 2.4032,
+      "step": 15554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.423215773396279e-05,
+      "loss": 1.0395,
+      "step": 15556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4227529389984265e-05,
+      "loss": 1.7913,
+      "step": 15558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.422290104600574e-05,
+      "loss": 0.0105,
+      "step": 15560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4218272702027216e-05,
+      "loss": 0.1428,
+      "step": 15562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4213644358048692e-05,
+      "loss": 0.2416,
+      "step": 15564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4209016014070167e-05,
+      "loss": 0.554,
+      "step": 15566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4204387670091642e-05,
+      "loss": 0.2534,
+      "step": 15568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4199759326113118e-05,
+      "loss": 1.5591,
+      "step": 15570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4195130982134593e-05,
+      "loss": 0.0018,
+      "step": 15572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.419050263815607e-05,
+      "loss": 4.878,
+      "step": 15574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4185874294177546e-05,
+      "loss": 2.474,
+      "step": 15576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4181245950199021e-05,
+      "loss": 1.6068,
+      "step": 15578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4176617606220493e-05,
+      "loss": 6.4687,
+      "step": 15580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4171989262241969e-05,
+      "loss": 4.7057,
+      "step": 15582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4167360918263444e-05,
+      "loss": 0.6187,
+      "step": 15584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.416273257428492e-05,
+      "loss": 3.5633,
+      "step": 15586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4158104230306397e-05,
+      "loss": 1.8648,
+      "step": 15588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4153475886327872e-05,
+      "loss": 1.1538,
+      "step": 15590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4148847542349348e-05,
+      "loss": 0.8771,
+      "step": 15592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4144219198370823e-05,
+      "loss": 0.0012,
+      "step": 15594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4139590854392298e-05,
+      "loss": 1.4604,
+      "step": 15596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4134962510413774e-05,
+      "loss": 0.063,
+      "step": 15598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.413033416643525e-05,
+      "loss": 0.1224,
+      "step": 15600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4125705822456725e-05,
+      "loss": 0.961,
+      "step": 15602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.41210774784782e-05,
+      "loss": 0.0003,
+      "step": 15604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4116449134499676e-05,
+      "loss": 0.7512,
+      "step": 15606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4111820790521153e-05,
+      "loss": 2.2854,
+      "step": 15608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4107192446542628e-05,
+      "loss": 0.7176,
+      "step": 15610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4102564102564104e-05,
+      "loss": 1.8438,
+      "step": 15612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4097935758585579e-05,
+      "loss": 2.9494,
+      "step": 15614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4093307414607054e-05,
+      "loss": 4.2931,
+      "step": 15616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.408867907062853e-05,
+      "loss": 0.005,
+      "step": 15618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4084050726650005e-05,
+      "loss": 0.001,
+      "step": 15620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.407942238267148e-05,
+      "loss": 0.8712,
+      "step": 15622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4074794038692956e-05,
+      "loss": 0.0006,
+      "step": 15624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4070165694714431e-05,
+      "loss": 0.0037,
+      "step": 15626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4065537350735909e-05,
+      "loss": 0.5691,
+      "step": 15628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4060909006757384e-05,
+      "loss": 4.5427,
+      "step": 15630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.405628066277886e-05,
+      "loss": 0.3646,
+      "step": 15632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4051652318800335e-05,
+      "loss": 1.7155,
+      "step": 15634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.404702397482181e-05,
+      "loss": 3.9707,
+      "step": 15636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4042395630843286e-05,
+      "loss": 2.9425,
+      "step": 15638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4037767286864761e-05,
+      "loss": 2.4522,
+      "step": 15640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4033138942886237e-05,
+      "loss": 2.0615,
+      "step": 15642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4028510598907712e-05,
+      "loss": 0.5035,
+      "step": 15644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4023882254929187e-05,
+      "loss": 0.0009,
+      "step": 15646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4019253910950665e-05,
+      "loss": 2.6186,
+      "step": 15648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.401462556697214e-05,
+      "loss": 0.8079,
+      "step": 15650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4009997222993612e-05,
+      "loss": 1.7268,
+      "step": 15652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4005368879015087e-05,
+      "loss": 6.8937,
+      "step": 15654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.4000740535036563e-05,
+      "loss": 2.0419,
+      "step": 15656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3996112191058038e-05,
+      "loss": 1.668,
+      "step": 15658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3991483847079515e-05,
+      "loss": 0.0003,
+      "step": 15660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.398685550310099e-05,
+      "loss": 0.6324,
+      "step": 15662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3982227159122466e-05,
+      "loss": 0.0005,
+      "step": 15664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3977598815143942e-05,
+      "loss": 4.147,
+      "step": 15666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3972970471165417e-05,
+      "loss": 0.6323,
+      "step": 15668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3968342127186893e-05,
+      "loss": 2.4483,
+      "step": 15670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3963713783208368e-05,
+      "loss": 0.088,
+      "step": 15672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3959085439229843e-05,
+      "loss": 0.2946,
+      "step": 15674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3954457095251319e-05,
+      "loss": 0.0005,
+      "step": 15676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3949828751272794e-05,
+      "loss": 0.0029,
+      "step": 15678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.394520040729427e-05,
+      "loss": 2.1261,
+      "step": 15680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3940572063315747e-05,
+      "loss": 0.333,
+      "step": 15682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3935943719337222e-05,
+      "loss": 0.0002,
+      "step": 15684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3931315375358698e-05,
+      "loss": 0.0002,
+      "step": 15686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3926687031380173e-05,
+      "loss": 0.0004,
+      "step": 15688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3922058687401648e-05,
+      "loss": 0.0811,
+      "step": 15690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3917430343423124e-05,
+      "loss": 3.4585,
+      "step": 15692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.39128019994446e-05,
+      "loss": 2.108,
+      "step": 15694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3908173655466075e-05,
+      "loss": 0.7527,
+      "step": 15696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.390354531148755e-05,
+      "loss": 1.3462,
+      "step": 15698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3898916967509026e-05,
+      "loss": 2.3154,
+      "step": 15700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3894288623530503e-05,
+      "loss": 0.5342,
+      "step": 15702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3889660279551978e-05,
+      "loss": 0.5448,
+      "step": 15704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3885031935573454e-05,
+      "loss": 0.0001,
+      "step": 15706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3880403591594929e-05,
+      "loss": 0.9911,
+      "step": 15708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3875775247616404e-05,
+      "loss": 1.6248,
+      "step": 15710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.387114690363788e-05,
+      "loss": 9.7773,
+      "step": 15712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3866518559659355e-05,
+      "loss": 0.0193,
+      "step": 15714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.386189021568083e-05,
+      "loss": 3.2518,
+      "step": 15716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3857261871702306e-05,
+      "loss": 0.0338,
+      "step": 15718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3852633527723782e-05,
+      "loss": 2.1929,
+      "step": 15720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3848005183745259e-05,
+      "loss": 0.0002,
+      "step": 15722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3843376839766734e-05,
+      "loss": 0.6903,
+      "step": 15724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3838748495788206e-05,
+      "loss": 1.8321,
+      "step": 15726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3834120151809682e-05,
+      "loss": 0.3929,
+      "step": 15728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3829491807831157e-05,
+      "loss": 1.7359,
+      "step": 15730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3824863463852632e-05,
+      "loss": 0.8298,
+      "step": 15732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.382023511987411e-05,
+      "loss": 3.5124,
+      "step": 15734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3815606775895585e-05,
+      "loss": 0.0003,
+      "step": 15736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.381097843191706e-05,
+      "loss": 5.7601,
+      "step": 15738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3806350087938536e-05,
+      "loss": 1.37,
+      "step": 15740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3801721743960011e-05,
+      "loss": 4.3003,
+      "step": 15742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3797093399981487e-05,
+      "loss": 1.6919,
+      "step": 15744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3792465056002962e-05,
+      "loss": 0.0002,
+      "step": 15746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3787836712024437e-05,
+      "loss": 0.2357,
+      "step": 15748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3783208368045913e-05,
+      "loss": 0.121,
+      "step": 15750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3778580024067388e-05,
+      "loss": 0.8663,
+      "step": 15752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3773951680088865e-05,
+      "loss": 0.0026,
+      "step": 15754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.376932333611034e-05,
+      "loss": 6.2195,
+      "step": 15756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3764694992131816e-05,
+      "loss": 3.873,
+      "step": 15758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3760066648153292e-05,
+      "loss": 2.7381,
+      "step": 15760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3755438304174767e-05,
+      "loss": 0.0004,
+      "step": 15762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3750809960196243e-05,
+      "loss": 1.738,
+      "step": 15764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3746181616217718e-05,
+      "loss": 0.1132,
+      "step": 15766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3741553272239193e-05,
+      "loss": 0.6325,
+      "step": 15768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3736924928260669e-05,
+      "loss": 2.3684,
+      "step": 15770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3732296584282144e-05,
+      "loss": 2.2323,
+      "step": 15772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3727668240303621e-05,
+      "loss": 0.0341,
+      "step": 15774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3723039896325097e-05,
+      "loss": 2.6818,
+      "step": 15776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3718411552346572e-05,
+      "loss": 5.4052,
+      "step": 15778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3713783208368048e-05,
+      "loss": 2.1789,
+      "step": 15780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3709154864389523e-05,
+      "loss": 5.1371,
+      "step": 15782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3704526520410998e-05,
+      "loss": 0.7577,
+      "step": 15784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3699898176432474e-05,
+      "loss": 0.0005,
+      "step": 15786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.369526983245395e-05,
+      "loss": 1.4599,
+      "step": 15788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3690641488475425e-05,
+      "loss": 0.001,
+      "step": 15790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.36860131444969e-05,
+      "loss": 1.1588,
+      "step": 15792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3681384800518377e-05,
+      "loss": 1.5208,
+      "step": 15794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3676756456539853e-05,
+      "loss": 0.0029,
+      "step": 15796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3672128112561325e-05,
+      "loss": 1.0548,
+      "step": 15798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.36674997685828e-05,
+      "loss": 0.6044,
+      "step": 15800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3662871424604276e-05,
+      "loss": 0.9302,
+      "step": 15802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3658243080625751e-05,
+      "loss": 3.4166,
+      "step": 15804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3653614736647226e-05,
+      "loss": 2.5832,
+      "step": 15806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3648986392668704e-05,
+      "loss": 0.0005,
+      "step": 15808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3644358048690179e-05,
+      "loss": 1.918,
+      "step": 15810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3639729704711654e-05,
+      "loss": 0.2688,
+      "step": 15812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.363510136073313e-05,
+      "loss": 1.7442,
+      "step": 15814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3630473016754605e-05,
+      "loss": 0.0752,
+      "step": 15816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.362584467277608e-05,
+      "loss": 1.0415,
+      "step": 15818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3621216328797556e-05,
+      "loss": 0.4182,
+      "step": 15820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3616587984819032e-05,
+      "loss": 5.0892,
+      "step": 15822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3611959640840507e-05,
+      "loss": 4.0952,
+      "step": 15824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3607331296861982e-05,
+      "loss": 5.4301,
+      "step": 15826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.360270295288346e-05,
+      "loss": 2.0956,
+      "step": 15828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3598074608904935e-05,
+      "loss": 8.2639,
+      "step": 15830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.359344626492641e-05,
+      "loss": 1.7535,
+      "step": 15832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3588817920947886e-05,
+      "loss": 2.8819,
+      "step": 15834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3584189576969361e-05,
+      "loss": 1.7211,
+      "step": 15836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3579561232990837e-05,
+      "loss": 0.15,
+      "step": 15838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3574932889012312e-05,
+      "loss": 0.7273,
+      "step": 15840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3570304545033787e-05,
+      "loss": 2.0509,
+      "step": 15842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3565676201055263e-05,
+      "loss": 1.164,
+      "step": 15844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3561047857076738e-05,
+      "loss": 2.1521,
+      "step": 15846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3556419513098215e-05,
+      "loss": 0.0058,
+      "step": 15848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3551791169119691e-05,
+      "loss": 0.1912,
+      "step": 15850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3547162825141166e-05,
+      "loss": 1.5257,
+      "step": 15852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3542534481162642e-05,
+      "loss": 1.0621,
+      "step": 15854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3537906137184117e-05,
+      "loss": 2.7447,
+      "step": 15856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3533277793205593e-05,
+      "loss": 0.5258,
+      "step": 15858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3528649449227068e-05,
+      "loss": 6.6669,
+      "step": 15860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3524021105248543e-05,
+      "loss": 2.4953,
+      "step": 15862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3519392761270019e-05,
+      "loss": 0.6499,
+      "step": 15864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3514764417291494e-05,
+      "loss": 0.0024,
+      "step": 15866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3510136073312971e-05,
+      "loss": 6.9857,
+      "step": 15868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3505507729334443e-05,
+      "loss": 2.1838,
+      "step": 15870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3500879385355919e-05,
+      "loss": 0.2249,
+      "step": 15872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3496251041377394e-05,
+      "loss": 1.8471,
+      "step": 15874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.349162269739887e-05,
+      "loss": 0.2182,
+      "step": 15876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3486994353420345e-05,
+      "loss": 2.1904,
+      "step": 15878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.348236600944182e-05,
+      "loss": 1.5118,
+      "step": 15880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3477737665463298e-05,
+      "loss": 0.0024,
+      "step": 15882
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3473109321484773e-05,
+      "loss": 2.283,
+      "step": 15884
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3468480977506249e-05,
+      "loss": 0.003,
+      "step": 15886
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3463852633527724e-05,
+      "loss": 4.2903,
+      "step": 15888
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.34592242895492e-05,
+      "loss": 3.3813,
+      "step": 15890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3454595945570675e-05,
+      "loss": 1.3725,
+      "step": 15892
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.344996760159215e-05,
+      "loss": 0.0016,
+      "step": 15894
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3445339257613626e-05,
+      "loss": 0.2818,
+      "step": 15896
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3440710913635101e-05,
+      "loss": 0.0524,
+      "step": 15898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3436082569656576e-05,
+      "loss": 0.0028,
+      "step": 15900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3431454225678054e-05,
+      "loss": 2.2432,
+      "step": 15902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3426825881699529e-05,
+      "loss": 1.8349,
+      "step": 15904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3422197537721004e-05,
+      "loss": 2.2411,
+      "step": 15906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.341756919374248e-05,
+      "loss": 1.285,
+      "step": 15908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3412940849763955e-05,
+      "loss": 4.3582,
+      "step": 15910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.340831250578543e-05,
+      "loss": 1.7554,
+      "step": 15912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3403684161806906e-05,
+      "loss": 3.1451,
+      "step": 15914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3399055817828382e-05,
+      "loss": 5.1497,
+      "step": 15916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3394427473849857e-05,
+      "loss": 0.0003,
+      "step": 15918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3389799129871332e-05,
+      "loss": 0.0222,
+      "step": 15920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.338517078589281e-05,
+      "loss": 0.4647,
+      "step": 15922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3380542441914285e-05,
+      "loss": 1.0441,
+      "step": 15924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.337591409793576e-05,
+      "loss": 3.6885,
+      "step": 15926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3371285753957236e-05,
+      "loss": 4.63,
+      "step": 15928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3366657409978711e-05,
+      "loss": 0.8465,
+      "step": 15930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3362029066000187e-05,
+      "loss": 7.1532,
+      "step": 15932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3357400722021662e-05,
+      "loss": 0.7371,
+      "step": 15934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3352772378043138e-05,
+      "loss": 1.3584,
+      "step": 15936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3348144034064613e-05,
+      "loss": 1.4629,
+      "step": 15938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3343515690086088e-05,
+      "loss": 1.2693,
+      "step": 15940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3338887346107562e-05,
+      "loss": 0.5137,
+      "step": 15942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3334259002129038e-05,
+      "loss": 1.9534,
+      "step": 15944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3329630658150513e-05,
+      "loss": 0.1781,
+      "step": 15946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3325002314171988e-05,
+      "loss": 0.0004,
+      "step": 15948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3320373970193464e-05,
+      "loss": 5.9152,
+      "step": 15950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.331574562621494e-05,
+      "loss": 0.0096,
+      "step": 15952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3311117282236416e-05,
+      "loss": 0.4117,
+      "step": 15954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3306488938257892e-05,
+      "loss": 3.2104,
+      "step": 15956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3301860594279367e-05,
+      "loss": 5.098,
+      "step": 15958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3297232250300843e-05,
+      "loss": 2.4288,
+      "step": 15960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3292603906322318e-05,
+      "loss": 0.0799,
+      "step": 15962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3287975562343793e-05,
+      "loss": 0.0017,
+      "step": 15964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3283347218365269e-05,
+      "loss": 1.1921,
+      "step": 15966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3278718874386744e-05,
+      "loss": 1.2666,
+      "step": 15968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.327409053040822e-05,
+      "loss": 0.185,
+      "step": 15970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3269462186429695e-05,
+      "loss": 1.2516,
+      "step": 15972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3264833842451172e-05,
+      "loss": 0.0017,
+      "step": 15974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3260205498472648e-05,
+      "loss": 2.2354,
+      "step": 15976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3255577154494123e-05,
+      "loss": 0.0082,
+      "step": 15978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3250948810515599e-05,
+      "loss": 0.1558,
+      "step": 15980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3246320466537074e-05,
+      "loss": 4.6608,
+      "step": 15982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.324169212255855e-05,
+      "loss": 5.5011,
+      "step": 15984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3237063778580025e-05,
+      "loss": 5.771,
+      "step": 15986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.32324354346015e-05,
+      "loss": 0.0231,
+      "step": 15988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3227807090622976e-05,
+      "loss": 0.2771,
+      "step": 15990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3223178746644451e-05,
+      "loss": 0.0056,
+      "step": 15992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3218550402665927e-05,
+      "loss": 2.3276,
+      "step": 15994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3213922058687404e-05,
+      "loss": 0.1362,
+      "step": 15996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3209293714708879e-05,
+      "loss": 3.909,
+      "step": 15998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3204665370730355e-05,
+      "loss": 0.0016,
+      "step": 16000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.320003702675183e-05,
+      "loss": 5.3176,
+      "step": 16002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3195408682773305e-05,
+      "loss": 5.2605,
+      "step": 16004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.319078033879478e-05,
+      "loss": 3.5027,
+      "step": 16006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3186151994816256e-05,
+      "loss": 2.2293,
+      "step": 16008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3181523650837732e-05,
+      "loss": 1.5591,
+      "step": 16010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3176895306859207e-05,
+      "loss": 0.6815,
+      "step": 16012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3172266962880682e-05,
+      "loss": 0.0021,
+      "step": 16014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3167638618902156e-05,
+      "loss": 8.9284,
+      "step": 16016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3163010274923632e-05,
+      "loss": 1.3165,
+      "step": 16018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3158381930945107e-05,
+      "loss": 1.7395,
+      "step": 16020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3153753586966582e-05,
+      "loss": 1.0839,
+      "step": 16022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3149125242988058e-05,
+      "loss": 1.2738,
+      "step": 16024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3144496899009533e-05,
+      "loss": 0.9717,
+      "step": 16026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.313986855503101e-05,
+      "loss": 0.6932,
+      "step": 16028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3135240211052486e-05,
+      "loss": 1.1069,
+      "step": 16030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3130611867073961e-05,
+      "loss": 3.4873,
+      "step": 16032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3125983523095437e-05,
+      "loss": 1.0881,
+      "step": 16034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3121355179116912e-05,
+      "loss": 1.9712,
+      "step": 16036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3116726835138388e-05,
+      "loss": 5.8077,
+      "step": 16038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3112098491159863e-05,
+      "loss": 1.0496,
+      "step": 16040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3107470147181338e-05,
+      "loss": 2.3881,
+      "step": 16042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3102841803202814e-05,
+      "loss": 1.4778,
+      "step": 16044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.309821345922429e-05,
+      "loss": 0.3003,
+      "step": 16046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3093585115245766e-05,
+      "loss": 0.0007,
+      "step": 16048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3088956771267242e-05,
+      "loss": 2.2471,
+      "step": 16050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3084328427288717e-05,
+      "loss": 1.6322,
+      "step": 16052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3079700083310193e-05,
+      "loss": 2.1509,
+      "step": 16054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3075071739331668e-05,
+      "loss": 1.1375,
+      "step": 16056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3070443395353144e-05,
+      "loss": 2.5298,
+      "step": 16058
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.3065815051374619e-05,
+      "loss": 1.3087,
+      "step": 16060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3061186707396094e-05,
+      "loss": 0.763,
+      "step": 16062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.305655836341757e-05,
+      "loss": 1.0048,
+      "step": 16064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3051930019439045e-05,
+      "loss": 2.2645,
+      "step": 16066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3047301675460522e-05,
+      "loss": 1.8442,
+      "step": 16068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3042673331481998e-05,
+      "loss": 0.5738,
+      "step": 16070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3038044987503473e-05,
+      "loss": 2.7756,
+      "step": 16072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3033416643524949e-05,
+      "loss": 1.6373,
+      "step": 16074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3028788299546424e-05,
+      "loss": 1.2451,
+      "step": 16076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.30241599555679e-05,
+      "loss": 2.0081,
+      "step": 16078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3019531611589375e-05,
+      "loss": 2.2573,
+      "step": 16080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.301490326761085e-05,
+      "loss": 1.9359,
+      "step": 16082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3010274923632326e-05,
+      "loss": 4.5271,
+      "step": 16084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3005646579653801e-05,
+      "loss": 3.0531,
+      "step": 16086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3001018235675275e-05,
+      "loss": 3.2231,
+      "step": 16088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.299638989169675e-05,
+      "loss": 6.7481,
+      "step": 16090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2991761547718226e-05,
+      "loss": 2.171,
+      "step": 16092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2987133203739701e-05,
+      "loss": 1.0035,
+      "step": 16094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2982504859761177e-05,
+      "loss": 2.182,
+      "step": 16096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2977876515782652e-05,
+      "loss": 1.3414,
+      "step": 16098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2973248171804127e-05,
+      "loss": 2.2192,
+      "step": 16100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2968619827825605e-05,
+      "loss": 0.0075,
+      "step": 16102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.296399148384708e-05,
+      "loss": 2.0304,
+      "step": 16104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2959363139868555e-05,
+      "loss": 3.2822,
+      "step": 16106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.295473479589003e-05,
+      "loss": 1.4108,
+      "step": 16108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2950106451911506e-05,
+      "loss": 2.1248,
+      "step": 16110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2945478107932982e-05,
+      "loss": 0.8374,
+      "step": 16112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2940849763954457e-05,
+      "loss": 1.8325,
+      "step": 16114
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2936221419975933e-05,
+      "loss": 0.182,
+      "step": 16116
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2931593075997408e-05,
+      "loss": 2.7985,
+      "step": 16118
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2926964732018883e-05,
+      "loss": 1.4231,
+      "step": 16120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.292233638804036e-05,
+      "loss": 0.8151,
+      "step": 16122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2917708044061836e-05,
+      "loss": 6.6907,
+      "step": 16124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2913079700083311e-05,
+      "loss": 1.3617,
+      "step": 16126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2908451356104787e-05,
+      "loss": 0.2355,
+      "step": 16128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2903823012126262e-05,
+      "loss": 1.4244,
+      "step": 16130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2899194668147738e-05,
+      "loss": 2.9212,
+      "step": 16132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2894566324169213e-05,
+      "loss": 1.1398,
+      "step": 16134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2889937980190688e-05,
+      "loss": 1.2627,
+      "step": 16136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2885309636212164e-05,
+      "loss": 3.4292,
+      "step": 16138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.288068129223364e-05,
+      "loss": 2.6377,
+      "step": 16140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2876052948255116e-05,
+      "loss": 0.0656,
+      "step": 16142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2871424604276592e-05,
+      "loss": 3.7439,
+      "step": 16144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2866796260298067e-05,
+      "loss": 1.91,
+      "step": 16146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2862167916319543e-05,
+      "loss": 2.9145,
+      "step": 16148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2857539572341018e-05,
+      "loss": 0.1616,
+      "step": 16150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2852911228362494e-05,
+      "loss": 0.6317,
+      "step": 16152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2848282884383969e-05,
+      "loss": 5.9546,
+      "step": 16154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2843654540405444e-05,
+      "loss": 0.132,
+      "step": 16156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.283902619642692e-05,
+      "loss": 0.1423,
+      "step": 16158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2834397852448394e-05,
+      "loss": 0.6536,
+      "step": 16160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2829769508469869e-05,
+      "loss": 0.0051,
+      "step": 16162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2825141164491344e-05,
+      "loss": 7.444,
+      "step": 16164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 4.8365,
+      "step": 16166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2815884476534295e-05,
+      "loss": 1.6826,
+      "step": 16168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.281125613255577e-05,
+      "loss": 0.382,
+      "step": 16170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2806627788577246e-05,
+      "loss": 5.7161,
+      "step": 16172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2801999444598722e-05,
+      "loss": 3.6571,
+      "step": 16174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2797371100620199e-05,
+      "loss": 4.0811,
+      "step": 16176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2792742756641674e-05,
+      "loss": 0.7795,
+      "step": 16178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.278811441266315e-05,
+      "loss": 2.6022,
+      "step": 16180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2783486068684625e-05,
+      "loss": 0.3845,
+      "step": 16182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.27788577247061e-05,
+      "loss": 5.0095,
+      "step": 16184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2774229380727576e-05,
+      "loss": 3.7153,
+      "step": 16186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2769601036749051e-05,
+      "loss": 0.0018,
+      "step": 16188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2764972692770527e-05,
+      "loss": 2.5138,
+      "step": 16190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2760344348792002e-05,
+      "loss": 0.0343,
+      "step": 16192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2755716004813477e-05,
+      "loss": 2.9664,
+      "step": 16194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2751087660834955e-05,
+      "loss": 2.5488,
+      "step": 16196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.274645931685643e-05,
+      "loss": 7.4165,
+      "step": 16198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2741830972877905e-05,
+      "loss": 1.2041,
+      "step": 16200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2737202628899381e-05,
+      "loss": 0.5951,
+      "step": 16202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2732574284920856e-05,
+      "loss": 5.3302,
+      "step": 16204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2727945940942332e-05,
+      "loss": 0.0444,
+      "step": 16206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2723317596963807e-05,
+      "loss": 0.0047,
+      "step": 16208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2718689252985283e-05,
+      "loss": 5.362,
+      "step": 16210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2714060909006758e-05,
+      "loss": 0.0131,
+      "step": 16212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2709432565028233e-05,
+      "loss": 5.2705,
+      "step": 16214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.270480422104971e-05,
+      "loss": 1.1378,
+      "step": 16216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2700175877071186e-05,
+      "loss": 0.0027,
+      "step": 16218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2695547533092661e-05,
+      "loss": 0.1544,
+      "step": 16220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2690919189114137e-05,
+      "loss": 2.0515,
+      "step": 16222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2686290845135612e-05,
+      "loss": 1.4813,
+      "step": 16224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2681662501157088e-05,
+      "loss": 0.4557,
+      "step": 16226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2677034157178563e-05,
+      "loss": 0.0069,
+      "step": 16228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2672405813200038e-05,
+      "loss": 0.3617,
+      "step": 16230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2667777469221512e-05,
+      "loss": 1.0313,
+      "step": 16232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2663149125242988e-05,
+      "loss": 0.0036,
+      "step": 16234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2658520781264463e-05,
+      "loss": 2.0536,
+      "step": 16236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2653892437285938e-05,
+      "loss": 0.709,
+      "step": 16238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2649264093307414e-05,
+      "loss": 1.2628,
+      "step": 16240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.264463574932889e-05,
+      "loss": 1.6647,
+      "step": 16242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2640007405350365e-05,
+      "loss": 0.0007,
+      "step": 16244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.263537906137184e-05,
+      "loss": 0.0006,
+      "step": 16246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2630750717393317e-05,
+      "loss": 0.4192,
+      "step": 16248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2626122373414793e-05,
+      "loss": 0.3129,
+      "step": 16250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2621494029436268e-05,
+      "loss": 2.0106,
+      "step": 16252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2616865685457744e-05,
+      "loss": 0.0012,
+      "step": 16254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2612237341479219e-05,
+      "loss": 0.0012,
+      "step": 16256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2607608997500694e-05,
+      "loss": 1.7085,
+      "step": 16258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.260298065352217e-05,
+      "loss": 1.1681,
+      "step": 16260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2598352309543645e-05,
+      "loss": 0.0013,
+      "step": 16262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.259372396556512e-05,
+      "loss": 0.6368,
+      "step": 16264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2589095621586596e-05,
+      "loss": 2.5544,
+      "step": 16266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2584467277608073e-05,
+      "loss": 1.3257,
+      "step": 16268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2579838933629549e-05,
+      "loss": 0.0749,
+      "step": 16270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2575210589651024e-05,
+      "loss": 2.3704,
+      "step": 16272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.25705822456725e-05,
+      "loss": 1.2646,
+      "step": 16274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2565953901693975e-05,
+      "loss": 3.0724,
+      "step": 16276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.256132555771545e-05,
+      "loss": 1.9161,
+      "step": 16278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2556697213736926e-05,
+      "loss": 1.2891,
+      "step": 16280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2552068869758401e-05,
+      "loss": 0.9119,
+      "step": 16282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2547440525779877e-05,
+      "loss": 0.0072,
+      "step": 16284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2542812181801352e-05,
+      "loss": 0.0004,
+      "step": 16286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.253818383782283e-05,
+      "loss": 1.7973,
+      "step": 16288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2533555493844305e-05,
+      "loss": 4.8028,
+      "step": 16290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.252892714986578e-05,
+      "loss": 1.171,
+      "step": 16292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2524298805887255e-05,
+      "loss": 3.5493,
+      "step": 16294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2519670461908731e-05,
+      "loss": 0.0005,
+      "step": 16296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2515042117930206e-05,
+      "loss": 0.453,
+      "step": 16298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2510413773951682e-05,
+      "loss": 0.3644,
+      "step": 16300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2505785429973157e-05,
+      "loss": 2.8214,
+      "step": 16302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2501157085994633e-05,
+      "loss": 2.7972,
+      "step": 16304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2496528742016108e-05,
+      "loss": 0.1817,
+      "step": 16306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2491900398037583e-05,
+      "loss": 0.717,
+      "step": 16308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2487272054059059e-05,
+      "loss": 2.1301,
+      "step": 16310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2482643710080534e-05,
+      "loss": 0.6788,
+      "step": 16312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.247801536610201e-05,
+      "loss": 2.0209,
+      "step": 16314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2473387022123485e-05,
+      "loss": 2.4101,
+      "step": 16316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.246875867814496e-05,
+      "loss": 1.9066,
+      "step": 16318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2464130334166436e-05,
+      "loss": 1.5182,
+      "step": 16320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2459501990187911e-05,
+      "loss": 0.3171,
+      "step": 16322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2454873646209387e-05,
+      "loss": 0.0004,
+      "step": 16324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2450245302230862e-05,
+      "loss": 1.8526,
+      "step": 16326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2445616958252338e-05,
+      "loss": 0.1933,
+      "step": 16328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2440988614273813e-05,
+      "loss": 0.7673,
+      "step": 16330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2436360270295289e-05,
+      "loss": 2.2558,
+      "step": 16332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2431731926316764e-05,
+      "loss": 2.5224,
+      "step": 16334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.242710358233824e-05,
+      "loss": 1.3792,
+      "step": 16336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2422475238359715e-05,
+      "loss": 1.6133,
+      "step": 16338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.241784689438119e-05,
+      "loss": 3.953,
+      "step": 16340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2413218550402667e-05,
+      "loss": 0.4873,
+      "step": 16342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2408590206424143e-05,
+      "loss": 0.3369,
+      "step": 16344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2403961862445618e-05,
+      "loss": 2.4693,
+      "step": 16346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2399333518467094e-05,
+      "loss": 3.9227,
+      "step": 16348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2394705174488569e-05,
+      "loss": 2.6552,
+      "step": 16350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2390076830510044e-05,
+      "loss": 3.5149,
+      "step": 16352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.238544848653152e-05,
+      "loss": 4.2598,
+      "step": 16354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2380820142552995e-05,
+      "loss": 0.0009,
+      "step": 16356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.237619179857447e-05,
+      "loss": 0.0039,
+      "step": 16358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2371563454595946e-05,
+      "loss": 2.0261,
+      "step": 16360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2366935110617422e-05,
+      "loss": 4.1267,
+      "step": 16362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2362306766638897e-05,
+      "loss": 0.93,
+      "step": 16364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2357678422660372e-05,
+      "loss": 2.5928,
+      "step": 16366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2353050078681848e-05,
+      "loss": 0.6581,
+      "step": 16368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2348421734703323e-05,
+      "loss": 0.3768,
+      "step": 16370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2343793390724799e-05,
+      "loss": 2.1599,
+      "step": 16372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2339165046746274e-05,
+      "loss": 1.7589,
+      "step": 16374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.233453670276775e-05,
+      "loss": 0.4627,
+      "step": 16376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2329908358789227e-05,
+      "loss": 0.0032,
+      "step": 16378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2325280014810702e-05,
+      "loss": 0.2507,
+      "step": 16380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2320651670832178e-05,
+      "loss": 0.5998,
+      "step": 16382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2316023326853653e-05,
+      "loss": 0.0202,
+      "step": 16384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2311394982875128e-05,
+      "loss": 0.8916,
+      "step": 16386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2306766638896604e-05,
+      "loss": 3.5778,
+      "step": 16388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.230213829491808e-05,
+      "loss": 3.7081,
+      "step": 16390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2297509950939555e-05,
+      "loss": 3.846,
+      "step": 16392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.229288160696103e-05,
+      "loss": 0.3014,
+      "step": 16394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2288253262982506e-05,
+      "loss": 0.0888,
+      "step": 16396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2283624919003981e-05,
+      "loss": 0.0003,
+      "step": 16398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2278996575025456e-05,
+      "loss": 0.0392,
+      "step": 16400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2274368231046932e-05,
+      "loss": 2.896,
+      "step": 16402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2269739887068407e-05,
+      "loss": 0.4299,
+      "step": 16404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2265111543089883e-05,
+      "loss": 0.1854,
+      "step": 16406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2260483199111358e-05,
+      "loss": 1.6227,
+      "step": 16408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2255854855132833e-05,
+      "loss": 5.167,
+      "step": 16410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2251226511154309e-05,
+      "loss": 0.8971,
+      "step": 16412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2246598167175784e-05,
+      "loss": 1.3134,
+      "step": 16414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2241969823197261e-05,
+      "loss": 3.2339,
+      "step": 16416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2237341479218737e-05,
+      "loss": 2.5472,
+      "step": 16418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2232713135240212e-05,
+      "loss": 1.8026,
+      "step": 16420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2228084791261688e-05,
+      "loss": 2.9392,
+      "step": 16422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2223456447283163e-05,
+      "loss": 0.8704,
+      "step": 16424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2218828103304639e-05,
+      "loss": 5.0245,
+      "step": 16426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2214199759326114e-05,
+      "loss": 0.3228,
+      "step": 16428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.220957141534759e-05,
+      "loss": 0.2864,
+      "step": 16430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2204943071369065e-05,
+      "loss": 2.5672,
+      "step": 16432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.220031472739054e-05,
+      "loss": 6.8949,
+      "step": 16434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2195686383412016e-05,
+      "loss": 0.4981,
+      "step": 16436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2191058039433491e-05,
+      "loss": 1.4934,
+      "step": 16438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2186429695454967e-05,
+      "loss": 2.1996,
+      "step": 16440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2181801351476442e-05,
+      "loss": 0.0487,
+      "step": 16442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2177173007497917e-05,
+      "loss": 1.8961,
+      "step": 16444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2172544663519393e-05,
+      "loss": 0.0873,
+      "step": 16446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2167916319540868e-05,
+      "loss": 2.9106,
+      "step": 16448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2163287975562344e-05,
+      "loss": 2.3383,
+      "step": 16450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.215865963158382e-05,
+      "loss": 0.4736,
+      "step": 16452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2154031287605296e-05,
+      "loss": 2.3421,
+      "step": 16454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2149402943626772e-05,
+      "loss": 4.6944,
+      "step": 16456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2144774599648247e-05,
+      "loss": 5.225,
+      "step": 16458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2140146255669722e-05,
+      "loss": 0.7766,
+      "step": 16460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2135517911691198e-05,
+      "loss": 0.3341,
+      "step": 16462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2130889567712673e-05,
+      "loss": 1.8148,
+      "step": 16464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2126261223734149e-05,
+      "loss": 1.7897,
+      "step": 16466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2121632879755624e-05,
+      "loss": 0.1112,
+      "step": 16468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.21170045357771e-05,
+      "loss": 1.1565,
+      "step": 16470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2112376191798575e-05,
+      "loss": 1.8882,
+      "step": 16472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.210774784782005e-05,
+      "loss": 2.0379,
+      "step": 16474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2103119503841526e-05,
+      "loss": 1.2443,
+      "step": 16476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2098491159863001e-05,
+      "loss": 2.5731,
+      "step": 16478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2093862815884477e-05,
+      "loss": 4.5285,
+      "step": 16480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2089234471905952e-05,
+      "loss": 0.2478,
+      "step": 16482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2084606127927428e-05,
+      "loss": 4.6885,
+      "step": 16484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2079977783948903e-05,
+      "loss": 1.5709,
+      "step": 16486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2075349439970378e-05,
+      "loss": 1.2734,
+      "step": 16488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2070721095991856e-05,
+      "loss": 1.1364,
+      "step": 16490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2066092752013331e-05,
+      "loss": 2.8797,
+      "step": 16492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2061464408034806e-05,
+      "loss": 5.6982,
+      "step": 16494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2056836064056282e-05,
+      "loss": 0.0048,
+      "step": 16496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2052207720077757e-05,
+      "loss": 0.0034,
+      "step": 16498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2047579376099233e-05,
+      "loss": 4.0809,
+      "step": 16500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2042951032120708e-05,
+      "loss": 0.7263,
+      "step": 16502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2038322688142182e-05,
+      "loss": 3.4813,
+      "step": 16504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2033694344163659e-05,
+      "loss": 0.0033,
+      "step": 16506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2029066000185134e-05,
+      "loss": 0.1044,
+      "step": 16508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.202443765620661e-05,
+      "loss": 0.1088,
+      "step": 16510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2019809312228085e-05,
+      "loss": 0.328,
+      "step": 16512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.201518096824956e-05,
+      "loss": 0.923,
+      "step": 16514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2010552624271036e-05,
+      "loss": 0.6846,
+      "step": 16516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2005924280292511e-05,
+      "loss": 1.1644,
+      "step": 16518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2001295936313987e-05,
+      "loss": 0.7502,
+      "step": 16520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1996667592335462e-05,
+      "loss": 1.7316,
+      "step": 16522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1992039248356938e-05,
+      "loss": 4.5402,
+      "step": 16524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1987410904378415e-05,
+      "loss": 2.7595,
+      "step": 16526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.198278256039989e-05,
+      "loss": 1.9295,
+      "step": 16528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1978154216421366e-05,
+      "loss": 0.8337,
+      "step": 16530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1973525872442841e-05,
+      "loss": 1.1085,
+      "step": 16532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1968897528464317e-05,
+      "loss": 3.8741,
+      "step": 16534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1964269184485792e-05,
+      "loss": 3.8946,
+      "step": 16536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1959640840507267e-05,
+      "loss": 0.8686,
+      "step": 16538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1955012496528743e-05,
+      "loss": 0.4987,
+      "step": 16540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1950384152550218e-05,
+      "loss": 0.1846,
+      "step": 16542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1945755808571694e-05,
+      "loss": 0.0581,
+      "step": 16544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1941127464593169e-05,
+      "loss": 0.0682,
+      "step": 16546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1936499120614645e-05,
+      "loss": 1.3629,
+      "step": 16548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.193187077663612e-05,
+      "loss": 2.2684,
+      "step": 16550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1927242432657595e-05,
+      "loss": 0.1936,
+      "step": 16552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.192261408867907e-05,
+      "loss": 0.338,
+      "step": 16554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1917985744700546e-05,
+      "loss": 1.2696,
+      "step": 16556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1913357400722022e-05,
+      "loss": 2.9663,
+      "step": 16558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1908729056743497e-05,
+      "loss": 0.5837,
+      "step": 16560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1904100712764974e-05,
+      "loss": 3.6227,
+      "step": 16562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.189947236878645e-05,
+      "loss": 2.5318,
+      "step": 16564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1894844024807925e-05,
+      "loss": 5.9708,
+      "step": 16566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.18902156808294e-05,
+      "loss": 0.0158,
+      "step": 16568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1885587336850876e-05,
+      "loss": 0.0574,
+      "step": 16570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1880958992872351e-05,
+      "loss": 0.003,
+      "step": 16572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1876330648893827e-05,
+      "loss": 4.6307,
+      "step": 16574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1871702304915302e-05,
+      "loss": 1.5728,
+      "step": 16576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1867073960936776e-05,
+      "loss": 0.3273,
+      "step": 16578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1862445616958253e-05,
+      "loss": 4.7165,
+      "step": 16580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1857817272979728e-05,
+      "loss": 0.3485,
+      "step": 16582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1853188929001204e-05,
+      "loss": 1.0407,
+      "step": 16584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.184856058502268e-05,
+      "loss": 1.8701,
+      "step": 16586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1843932241044155e-05,
+      "loss": 0.0249,
+      "step": 16588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.183930389706563e-05,
+      "loss": 0.5137,
+      "step": 16590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1834675553087106e-05,
+      "loss": 1.0001,
+      "step": 16592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1830047209108581e-05,
+      "loss": 0.0262,
+      "step": 16594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1825418865130056e-05,
+      "loss": 0.505,
+      "step": 16596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1820790521151532e-05,
+      "loss": 1.4374,
+      "step": 16598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1816162177173009e-05,
+      "loss": 0.1597,
+      "step": 16600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1811533833194484e-05,
+      "loss": 1.433,
+      "step": 16602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.180690548921596e-05,
+      "loss": 0.0011,
+      "step": 16604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1802277145237435e-05,
+      "loss": 4.9954,
+      "step": 16606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.179764880125891e-05,
+      "loss": 0.0657,
+      "step": 16608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1793020457280386e-05,
+      "loss": 0.444,
+      "step": 16610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1788392113301862e-05,
+      "loss": 0.1062,
+      "step": 16612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1783763769323335e-05,
+      "loss": 0.0892,
+      "step": 16614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1779135425344812e-05,
+      "loss": 4.7708,
+      "step": 16616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1774507081366288e-05,
+      "loss": 5.361,
+      "step": 16618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1769878737387763e-05,
+      "loss": 0.2873,
+      "step": 16620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1765250393409239e-05,
+      "loss": 0.9596,
+      "step": 16622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1760622049430714e-05,
+      "loss": 0.012,
+      "step": 16624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.175599370545219e-05,
+      "loss": 0.0078,
+      "step": 16626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1751365361473665e-05,
+      "loss": 0.4386,
+      "step": 16628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.174673701749514e-05,
+      "loss": 4.1576,
+      "step": 16630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1742108673516616e-05,
+      "loss": 1.8449,
+      "step": 16632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1737480329538091e-05,
+      "loss": 0.5614,
+      "step": 16634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1732851985559568e-05,
+      "loss": 5.4648,
+      "step": 16636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1728223641581044e-05,
+      "loss": 1.1665,
+      "step": 16638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.172359529760252e-05,
+      "loss": 0.0337,
+      "step": 16640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1718966953623995e-05,
+      "loss": 2.5813,
+      "step": 16642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.171433860964547e-05,
+      "loss": 1.9368,
+      "step": 16644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1709710265666945e-05,
+      "loss": 0.0087,
+      "step": 16646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1705081921688421e-05,
+      "loss": 0.7183,
+      "step": 16648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1700453577709895e-05,
+      "loss": 7.2425,
+      "step": 16650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1695825233731372e-05,
+      "loss": 1.0771,
+      "step": 16652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1691196889752847e-05,
+      "loss": 1.4651,
+      "step": 16654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1686568545774323e-05,
+      "loss": 2.6842,
+      "step": 16656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1681940201795798e-05,
+      "loss": 2.3543,
+      "step": 16658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1677311857817273e-05,
+      "loss": 0.2186,
+      "step": 16660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1672683513838749e-05,
+      "loss": 0.2574,
+      "step": 16662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1668055169860224e-05,
+      "loss": 0.0228,
+      "step": 16664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.16634268258817e-05,
+      "loss": 1.4801,
+      "step": 16666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1658798481903175e-05,
+      "loss": 2.8312,
+      "step": 16668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.165417013792465e-05,
+      "loss": 2.5087,
+      "step": 16670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1649541793946128e-05,
+      "loss": 0.0004,
+      "step": 16672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1644913449967603e-05,
+      "loss": 0.6064,
+      "step": 16674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1640285105989079e-05,
+      "loss": 2.7847,
+      "step": 16676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1635656762010554e-05,
+      "loss": 1.2125,
+      "step": 16678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.163102841803203e-05,
+      "loss": 0.8196,
+      "step": 16680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1626400074053505e-05,
+      "loss": 0.2525,
+      "step": 16682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.162177173007498e-05,
+      "loss": 0.1103,
+      "step": 16684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1617143386096454e-05,
+      "loss": 0.5634,
+      "step": 16686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.161251504211793e-05,
+      "loss": 0.007,
+      "step": 16688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1607886698139406e-05,
+      "loss": 0.7245,
+      "step": 16690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1603258354160882e-05,
+      "loss": 3.683,
+      "step": 16692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1598630010182357e-05,
+      "loss": 0.0213,
+      "step": 16694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1594001666203833e-05,
+      "loss": 0.47,
+      "step": 16696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1589373322225308e-05,
+      "loss": 6.8546,
+      "step": 16698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1584744978246784e-05,
+      "loss": 2.8546,
+      "step": 16700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1580116634268259e-05,
+      "loss": 0.1099,
+      "step": 16702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1575488290289734e-05,
+      "loss": 3.5163,
+      "step": 16704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.157085994631121e-05,
+      "loss": 2.7102,
+      "step": 16706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1566231602332685e-05,
+      "loss": 1.1725,
+      "step": 16708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1561603258354162e-05,
+      "loss": 1.6796,
+      "step": 16710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1556974914375638e-05,
+      "loss": 1.7868,
+      "step": 16712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1552346570397113e-05,
+      "loss": 0.6459,
+      "step": 16714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1547718226418589e-05,
+      "loss": 0.0272,
+      "step": 16716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1543089882440064e-05,
+      "loss": 0.1399,
+      "step": 16718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.153846153846154e-05,
+      "loss": 3.499,
+      "step": 16720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1533833194483013e-05,
+      "loss": 3.0178,
+      "step": 16722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1529204850504489e-05,
+      "loss": 0.1781,
+      "step": 16724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1524576506525966e-05,
+      "loss": 5.3839,
+      "step": 16726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1519948162547441e-05,
+      "loss": 4.1366,
+      "step": 16728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1515319818568917e-05,
+      "loss": 0.035,
+      "step": 16730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1510691474590392e-05,
+      "loss": 3.4906,
+      "step": 16732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1506063130611868e-05,
+      "loss": 0.1058,
+      "step": 16734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1501434786633343e-05,
+      "loss": 0.0003,
+      "step": 16736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1496806442654818e-05,
+      "loss": 2.1523,
+      "step": 16738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1492178098676294e-05,
+      "loss": 5.2351,
+      "step": 16740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.148754975469777e-05,
+      "loss": 1.0037,
+      "step": 16742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1482921410719245e-05,
+      "loss": 0.0064,
+      "step": 16744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1478293066740722e-05,
+      "loss": 0.0008,
+      "step": 16746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1473664722762197e-05,
+      "loss": 6.8848,
+      "step": 16748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1469036378783673e-05,
+      "loss": 1.3045,
+      "step": 16750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1464408034805148e-05,
+      "loss": 0.7471,
+      "step": 16752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1459779690826623e-05,
+      "loss": 2.5351,
+      "step": 16754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1455151346848099e-05,
+      "loss": 0.2926,
+      "step": 16756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1450523002869573e-05,
+      "loss": 1.9691,
+      "step": 16758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1445894658891048e-05,
+      "loss": 2.5232,
+      "step": 16760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1441266314912525e-05,
+      "loss": 2.3937,
+      "step": 16762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1436637970934e-05,
+      "loss": 0.4834,
+      "step": 16764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1432009626955476e-05,
+      "loss": 3.8064,
+      "step": 16766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1427381282976951e-05,
+      "loss": 1.5588,
+      "step": 16768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1422752938998427e-05,
+      "loss": 0.835,
+      "step": 16770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1418124595019902e-05,
+      "loss": 4.7609,
+      "step": 16772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1413496251041378e-05,
+      "loss": 1.2548,
+      "step": 16774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1408867907062853e-05,
+      "loss": 0.0192,
+      "step": 16776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1404239563084329e-05,
+      "loss": 0.2838,
+      "step": 16778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1399611219105804e-05,
+      "loss": 5.571,
+      "step": 16780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1394982875127281e-05,
+      "loss": 0.0974,
+      "step": 16782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1390354531148757e-05,
+      "loss": 3.1831,
+      "step": 16784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1385726187170232e-05,
+      "loss": 4.732,
+      "step": 16786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1381097843191707e-05,
+      "loss": 5.1802,
+      "step": 16788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1376469499213183e-05,
+      "loss": 1.9624,
+      "step": 16790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1371841155234658e-05,
+      "loss": 0.2076,
+      "step": 16792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1367212811256132e-05,
+      "loss": 1.2407,
+      "step": 16794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1362584467277607e-05,
+      "loss": 4.5882,
+      "step": 16796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1357956123299083e-05,
+      "loss": 1.9915,
+      "step": 16798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.135332777932056e-05,
+      "loss": 0.069,
+      "step": 16800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1348699435342035e-05,
+      "loss": 6.3115,
+      "step": 16802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.134407109136351e-05,
+      "loss": 3.0309,
+      "step": 16804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1339442747384986e-05,
+      "loss": 4.2865,
+      "step": 16806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1334814403406462e-05,
+      "loss": 1.5716,
+      "step": 16808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1330186059427937e-05,
+      "loss": 1.7851,
+      "step": 16810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1325557715449412e-05,
+      "loss": 2.1949,
+      "step": 16812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1320929371470888e-05,
+      "loss": 3.8254,
+      "step": 16814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1316301027492363e-05,
+      "loss": 2.1017,
+      "step": 16816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1311672683513839e-05,
+      "loss": 1.881,
+      "step": 16818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1307044339535316e-05,
+      "loss": 1.5557,
+      "step": 16820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1302415995556791e-05,
+      "loss": 2.0854,
+      "step": 16822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1297787651578267e-05,
+      "loss": 0.8488,
+      "step": 16824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1293159307599742e-05,
+      "loss": 4.9022,
+      "step": 16826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1288530963621218e-05,
+      "loss": 0.7587,
+      "step": 16828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1283902619642693e-05,
+      "loss": 2.4087,
+      "step": 16830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1279274275664167e-05,
+      "loss": 1.3881,
+      "step": 16832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1274645931685642e-05,
+      "loss": 1.7943,
+      "step": 16834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.127001758770712e-05,
+      "loss": 2.0763,
+      "step": 16836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1265389243728595e-05,
+      "loss": 0.1258,
+      "step": 16838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.126076089975007e-05,
+      "loss": 1.7021,
+      "step": 16840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1256132555771546e-05,
+      "loss": 7.253,
+      "step": 16842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1251504211793021e-05,
+      "loss": 0.2867,
+      "step": 16844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1246875867814496e-05,
+      "loss": 0.3349,
+      "step": 16846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1242247523835972e-05,
+      "loss": 1.7549,
+      "step": 16848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1237619179857447e-05,
+      "loss": 3.2753,
+      "step": 16850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1232990835878923e-05,
+      "loss": 0.6607,
+      "step": 16852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1228362491900398e-05,
+      "loss": 1.084,
+      "step": 16854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1223734147921875e-05,
+      "loss": 0.413,
+      "step": 16856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.121910580394335e-05,
+      "loss": 3.0113,
+      "step": 16858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1214477459964826e-05,
+      "loss": 0.1664,
+      "step": 16860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1209849115986301e-05,
+      "loss": 0.4163,
+      "step": 16862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1205220772007777e-05,
+      "loss": 3.4949,
+      "step": 16864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1200592428029252e-05,
+      "loss": 0.5388,
+      "step": 16866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1195964084050726e-05,
+      "loss": 2.1036,
+      "step": 16868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1191335740072201e-05,
+      "loss": 0.6572,
+      "step": 16870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1186707396093679e-05,
+      "loss": 0.6285,
+      "step": 16872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1182079052115154e-05,
+      "loss": 1.5634,
+      "step": 16874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.117745070813663e-05,
+      "loss": 3.8342,
+      "step": 16876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1172822364158105e-05,
+      "loss": 3.3739,
+      "step": 16878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.116819402017958e-05,
+      "loss": 0.2247,
+      "step": 16880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1163565676201056e-05,
+      "loss": 2.1792,
+      "step": 16882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1158937332222531e-05,
+      "loss": 4.7664,
+      "step": 16884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1154308988244007e-05,
+      "loss": 0.0346,
+      "step": 16886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1149680644265482e-05,
+      "loss": 2.2282,
+      "step": 16888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1145052300286957e-05,
+      "loss": 6.1701,
+      "step": 16890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1140423956308433e-05,
+      "loss": 1.5605,
+      "step": 16892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.113579561232991e-05,
+      "loss": 0.2654,
+      "step": 16894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1131167268351385e-05,
+      "loss": 2.2673,
+      "step": 16896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.112653892437286e-05,
+      "loss": 0.0198,
+      "step": 16898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1121910580394336e-05,
+      "loss": 3.0247,
+      "step": 16900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1117282236415812e-05,
+      "loss": 4.1963,
+      "step": 16902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1112653892437285e-05,
+      "loss": 0.443,
+      "step": 16904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.110802554845876e-05,
+      "loss": 1.8901,
+      "step": 16906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1103397204480236e-05,
+      "loss": 0.8102,
+      "step": 16908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1098768860501713e-05,
+      "loss": 1.529,
+      "step": 16910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1094140516523189e-05,
+      "loss": 0.1012,
+      "step": 16912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1089512172544664e-05,
+      "loss": 4.1772,
+      "step": 16914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.108488382856614e-05,
+      "loss": 0.0005,
+      "step": 16916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1080255484587615e-05,
+      "loss": 0.7784,
+      "step": 16918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.107562714060909e-05,
+      "loss": 0.1371,
+      "step": 16920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1070998796630566e-05,
+      "loss": 8.0062,
+      "step": 16922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1066370452652041e-05,
+      "loss": 0.2997,
+      "step": 16924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1061742108673517e-05,
+      "loss": 2.1511,
+      "step": 16926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.1057113764694992e-05,
+      "loss": 4.0759,
+      "step": 16928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.105248542071647e-05,
+      "loss": 2.0988,
+      "step": 16930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1047857076737945e-05,
+      "loss": 0.191,
+      "step": 16932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.104322873275942e-05,
+      "loss": 0.0869,
+      "step": 16934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1038600388780896e-05,
+      "loss": 0.3508,
+      "step": 16936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1033972044802371e-05,
+      "loss": 1.0143,
+      "step": 16938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1029343700823845e-05,
+      "loss": 0.0005,
+      "step": 16940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.102471535684532e-05,
+      "loss": 6.1118,
+      "step": 16942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1020087012866796e-05,
+      "loss": 1.1614,
+      "step": 16944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1015458668888273e-05,
+      "loss": 7.184,
+      "step": 16946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1010830324909748e-05,
+      "loss": 1.3113,
+      "step": 16948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1006201980931224e-05,
+      "loss": 1.1853,
+      "step": 16950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.1001573636952699e-05,
+      "loss": 2.208,
+      "step": 16952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0996945292974174e-05,
+      "loss": 0.1307,
+      "step": 16954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.099231694899565e-05,
+      "loss": 2.4308,
+      "step": 16956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0987688605017125e-05,
+      "loss": 4.2964,
+      "step": 16958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.09830602610386e-05,
+      "loss": 1.2497,
+      "step": 16960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0978431917060076e-05,
+      "loss": 2.1841,
+      "step": 16962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0973803573081551e-05,
+      "loss": 0.4181,
+      "step": 16964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0969175229103029e-05,
+      "loss": 0.0009,
+      "step": 16966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0964546885124504e-05,
+      "loss": 1.4341,
+      "step": 16968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.095991854114598e-05,
+      "loss": 1.075,
+      "step": 16970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0955290197167455e-05,
+      "loss": 0.2607,
+      "step": 16972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.095066185318893e-05,
+      "loss": 0.537,
+      "step": 16974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0946033509210404e-05,
+      "loss": 0.0009,
+      "step": 16976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.094140516523188e-05,
+      "loss": 0.442,
+      "step": 16978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0936776821253355e-05,
+      "loss": 2.023,
+      "step": 16980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.093214847727483e-05,
+      "loss": 0.8847,
+      "step": 16982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0927520133296307e-05,
+      "loss": 0.6252,
+      "step": 16984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0922891789317783e-05,
+      "loss": 0.1617,
+      "step": 16986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0918263445339258e-05,
+      "loss": 2.3594,
+      "step": 16988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0913635101360734e-05,
+      "loss": 0.6537,
+      "step": 16990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0909006757382209e-05,
+      "loss": 0.0066,
+      "step": 16992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0904378413403685e-05,
+      "loss": 4.1276,
+      "step": 16994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.089975006942516e-05,
+      "loss": 1.2777,
+      "step": 16996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0895121725446635e-05,
+      "loss": 2.524,
+      "step": 16998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.089049338146811e-05,
+      "loss": 5.4557,
+      "step": 17000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0885865037489586e-05,
+      "loss": 5.2187,
+      "step": 17002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0881236693511063e-05,
+      "loss": 0.2405,
+      "step": 17004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0876608349532539e-05,
+      "loss": 0.1773,
+      "step": 17006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0871980005554014e-05,
+      "loss": 0.0012,
+      "step": 17008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.086735166157549e-05,
+      "loss": 0.0013,
+      "step": 17010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0862723317596963e-05,
+      "loss": 3.792,
+      "step": 17012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0858094973618439e-05,
+      "loss": 0.4929,
+      "step": 17014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0853466629639914e-05,
+      "loss": 0.297,
+      "step": 17016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.084883828566139e-05,
+      "loss": 4.3301,
+      "step": 17018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0844209941682867e-05,
+      "loss": 0.416,
+      "step": 17020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0839581597704342e-05,
+      "loss": 4.7001,
+      "step": 17022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0834953253725818e-05,
+      "loss": 3.4441,
+      "step": 17024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0830324909747293e-05,
+      "loss": 2.8114,
+      "step": 17026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0825696565768768e-05,
+      "loss": 1.7623,
+      "step": 17028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0821068221790244e-05,
+      "loss": 2.2914,
+      "step": 17030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.081643987781172e-05,
+      "loss": 0.8645,
+      "step": 17032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0811811533833195e-05,
+      "loss": 0.6164,
+      "step": 17034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.080718318985467e-05,
+      "loss": 0.0023,
+      "step": 17036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0802554845876146e-05,
+      "loss": 1.8774,
+      "step": 17038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0797926501897623e-05,
+      "loss": 2.6033,
+      "step": 17040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0793298157919098e-05,
+      "loss": 1.7116,
+      "step": 17042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0788669813940574e-05,
+      "loss": 2.423,
+      "step": 17044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0784041469962049e-05,
+      "loss": 0.7875,
+      "step": 17046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0779413125983523e-05,
+      "loss": 0.4087,
+      "step": 17048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0774784782004998e-05,
+      "loss": 0.6632,
+      "step": 17050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0770156438026474e-05,
+      "loss": 1.0605,
+      "step": 17052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0765528094047949e-05,
+      "loss": 0.3403,
+      "step": 17054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0760899750069426e-05,
+      "loss": 0.8363,
+      "step": 17056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0756271406090902e-05,
+      "loss": 0.1571,
+      "step": 17058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0751643062112377e-05,
+      "loss": 0.0029,
+      "step": 17060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0747014718133852e-05,
+      "loss": 1.2687,
+      "step": 17062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0742386374155328e-05,
+      "loss": 0.0008,
+      "step": 17064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0737758030176803e-05,
+      "loss": 1.9382,
+      "step": 17066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0733129686198279e-05,
+      "loss": 2.4272,
+      "step": 17068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0728501342219754e-05,
+      "loss": 2.7795,
+      "step": 17070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.072387299824123e-05,
+      "loss": 3.1202,
+      "step": 17072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0719244654262705e-05,
+      "loss": 2.2384,
+      "step": 17074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0714616310284182e-05,
+      "loss": 0.0015,
+      "step": 17076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0709987966305657e-05,
+      "loss": 0.0648,
+      "step": 17078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0705359622327133e-05,
+      "loss": 0.1727,
+      "step": 17080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0700731278348608e-05,
+      "loss": 2.861,
+      "step": 17082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0696102934370082e-05,
+      "loss": 0.4665,
+      "step": 17084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0691474590391557e-05,
+      "loss": 2.7212,
+      "step": 17086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0686846246413033e-05,
+      "loss": 0.5035,
+      "step": 17088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0682217902434508e-05,
+      "loss": 0.0019,
+      "step": 17090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0677589558455984e-05,
+      "loss": 1.9052,
+      "step": 17092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0672961214477461e-05,
+      "loss": 0.1834,
+      "step": 17094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0668332870498936e-05,
+      "loss": 0.0013,
+      "step": 17096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0663704526520412e-05,
+      "loss": 0.4291,
+      "step": 17098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0659076182541887e-05,
+      "loss": 2.5832,
+      "step": 17100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0654447838563363e-05,
+      "loss": 0.0298,
+      "step": 17102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0649819494584838e-05,
+      "loss": 0.0005,
+      "step": 17104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0645191150606313e-05,
+      "loss": 4.1944,
+      "step": 17106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0640562806627789e-05,
+      "loss": 0.1975,
+      "step": 17108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0635934462649264e-05,
+      "loss": 0.1438,
+      "step": 17110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.063130611867074e-05,
+      "loss": 2.902,
+      "step": 17112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0626677774692217e-05,
+      "loss": 1.1331,
+      "step": 17114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0622049430713692e-05,
+      "loss": 1.2102,
+      "step": 17116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0617421086735168e-05,
+      "loss": 0.0052,
+      "step": 17118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0612792742756643e-05,
+      "loss": 2.6528,
+      "step": 17120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0608164398778117e-05,
+      "loss": 2.3246,
+      "step": 17122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0603536054799592e-05,
+      "loss": 2.2081,
+      "step": 17124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0598907710821068e-05,
+      "loss": 1.8387,
+      "step": 17126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0594279366842543e-05,
+      "loss": 0.0257,
+      "step": 17128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.058965102286402e-05,
+      "loss": 1.2738,
+      "step": 17130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0585022678885496e-05,
+      "loss": 0.3273,
+      "step": 17132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0580394334906971e-05,
+      "loss": 1.6029,
+      "step": 17134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0575765990928446e-05,
+      "loss": 2.624,
+      "step": 17136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0571137646949922e-05,
+      "loss": 3.7664,
+      "step": 17138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0566509302971397e-05,
+      "loss": 0.0021,
+      "step": 17140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0561880958992873e-05,
+      "loss": 1.2749,
+      "step": 17142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0557252615014348e-05,
+      "loss": 0.7057,
+      "step": 17144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0552624271035824e-05,
+      "loss": 0.5493,
+      "step": 17146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0547995927057299e-05,
+      "loss": 0.2028,
+      "step": 17148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0543367583078776e-05,
+      "loss": 0.0337,
+      "step": 17150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0538739239100252e-05,
+      "loss": 4.1301,
+      "step": 17152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0534110895121727e-05,
+      "loss": 2.1113,
+      "step": 17154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0529482551143202e-05,
+      "loss": 3.1187,
+      "step": 17156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0524854207164676e-05,
+      "loss": 2.9963,
+      "step": 17158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0520225863186152e-05,
+      "loss": 8.4848,
+      "step": 17160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0515597519207627e-05,
+      "loss": 0.0461,
+      "step": 17162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0510969175229102e-05,
+      "loss": 0.0011,
+      "step": 17164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.050634083125058e-05,
+      "loss": 1.894,
+      "step": 17166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0501712487272055e-05,
+      "loss": 0.0844,
+      "step": 17168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.049708414329353e-05,
+      "loss": 1.6824,
+      "step": 17170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0492455799315006e-05,
+      "loss": 0.7423,
+      "step": 17172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0487827455336481e-05,
+      "loss": 0.0345,
+      "step": 17174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0483199111357957e-05,
+      "loss": 0.4194,
+      "step": 17176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0478570767379432e-05,
+      "loss": 0.5926,
+      "step": 17178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0473942423400908e-05,
+      "loss": 0.4157,
+      "step": 17180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0469314079422383e-05,
+      "loss": 1.061,
+      "step": 17182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0464685735443858e-05,
+      "loss": 4.5157,
+      "step": 17184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0460057391465335e-05,
+      "loss": 4.4674,
+      "step": 17186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0455429047486811e-05,
+      "loss": 0.1391,
+      "step": 17188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0450800703508286e-05,
+      "loss": 4.2405,
+      "step": 17190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0446172359529762e-05,
+      "loss": 2.2707,
+      "step": 17192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0441544015551235e-05,
+      "loss": 0.8526,
+      "step": 17194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0436915671572711e-05,
+      "loss": 0.9887,
+      "step": 17196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0432287327594186e-05,
+      "loss": 0.1156,
+      "step": 17198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0427658983615662e-05,
+      "loss": 1.1315,
+      "step": 17200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0423030639637137e-05,
+      "loss": 5.5751,
+      "step": 17202
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0418402295658614e-05,
+      "loss": 0.9999,
+      "step": 17204
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.041377395168009e-05,
+      "loss": 4.0054,
+      "step": 17206
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0409145607701565e-05,
+      "loss": 1.5272,
+      "step": 17208
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.040451726372304e-05,
+      "loss": 0.004,
+      "step": 17210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0399888919744516e-05,
+      "loss": 4.2545,
+      "step": 17212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0395260575765991e-05,
+      "loss": 1.2267,
+      "step": 17214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0390632231787467e-05,
+      "loss": 0.0027,
+      "step": 17216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0386003887808942e-05,
+      "loss": 2.9734,
+      "step": 17218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0381375543830418e-05,
+      "loss": 4.8909,
+      "step": 17220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0376747199851893e-05,
+      "loss": 3.7966,
+      "step": 17222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.037211885587337e-05,
+      "loss": 0.9652,
+      "step": 17224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0367490511894846e-05,
+      "loss": 0.0185,
+      "step": 17226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0362862167916321e-05,
+      "loss": 1.3028,
+      "step": 17228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0358233823937795e-05,
+      "loss": 0.0013,
+      "step": 17230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.035360547995927e-05,
+      "loss": 0.0331,
+      "step": 17232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0348977135980746e-05,
+      "loss": 1.0373,
+      "step": 17234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0344348792002221e-05,
+      "loss": 1.5519,
+      "step": 17236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0339720448023697e-05,
+      "loss": 0.1021,
+      "step": 17238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0335092104045174e-05,
+      "loss": 2.202,
+      "step": 17240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0330463760066649e-05,
+      "loss": 3.0737,
+      "step": 17242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0325835416088124e-05,
+      "loss": 2.0077,
+      "step": 17244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.03212070721096e-05,
+      "loss": 1.53,
+      "step": 17246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0316578728131075e-05,
+      "loss": 0.0721,
+      "step": 17248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.031195038415255e-05,
+      "loss": 1.406,
+      "step": 17250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0307322040174026e-05,
+      "loss": 0.0992,
+      "step": 17252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0302693696195502e-05,
+      "loss": 2.1148,
+      "step": 17254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0298065352216977e-05,
+      "loss": 2.6455,
+      "step": 17256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0293437008238452e-05,
+      "loss": 2.1695,
+      "step": 17258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.028880866425993e-05,
+      "loss": 0.1518,
+      "step": 17260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0284180320281405e-05,
+      "loss": 0.7541,
+      "step": 17262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.027955197630288e-05,
+      "loss": 0.2293,
+      "step": 17264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0274923632324354e-05,
+      "loss": 2.4381,
+      "step": 17266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.027029528834583e-05,
+      "loss": 0.0022,
+      "step": 17268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0265666944367305e-05,
+      "loss": 1.4766,
+      "step": 17270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.026103860038878e-05,
+      "loss": 2.0336,
+      "step": 17272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.8232,
+      "step": 17274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0251781912431733e-05,
+      "loss": 0.0052,
+      "step": 17276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0247153568453208e-05,
+      "loss": 4.247,
+      "step": 17278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0242525224474684e-05,
+      "loss": 7.6466,
+      "step": 17280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.023789688049616e-05,
+      "loss": 0.3169,
+      "step": 17282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0233268536517635e-05,
+      "loss": 2.7099,
+      "step": 17284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.022864019253911e-05,
+      "loss": 1.4207,
+      "step": 17286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0224011848560586e-05,
+      "loss": 1.4546,
+      "step": 17288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0219383504582061e-05,
+      "loss": 1.584,
+      "step": 17290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0214755160603536e-05,
+      "loss": 7.1214,
+      "step": 17292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0210126816625012e-05,
+      "loss": 0.0042,
+      "step": 17294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0205498472646487e-05,
+      "loss": 4.1859,
+      "step": 17296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0200870128667964e-05,
+      "loss": 0.0013,
+      "step": 17298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.019624178468944e-05,
+      "loss": 0.0618,
+      "step": 17300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0191613440710913e-05,
+      "loss": 0.0316,
+      "step": 17302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0186985096732389e-05,
+      "loss": 1.8975,
+      "step": 17304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0182356752753864e-05,
+      "loss": 1.4024,
+      "step": 17306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.017772840877534e-05,
+      "loss": 3.5665,
+      "step": 17308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0173100064796815e-05,
+      "loss": 2.2342,
+      "step": 17310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.016847172081829e-05,
+      "loss": 3.6465,
+      "step": 17312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0163843376839768e-05,
+      "loss": 0.5749,
+      "step": 17314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0159215032861243e-05,
+      "loss": 2.5442,
+      "step": 17316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0154586688882719e-05,
+      "loss": 2.3336,
+      "step": 17318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0149958344904194e-05,
+      "loss": 0.3227,
+      "step": 17320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.014533000092567e-05,
+      "loss": 3.0819,
+      "step": 17322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0140701656947145e-05,
+      "loss": 3.42,
+      "step": 17324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.013607331296862e-05,
+      "loss": 1.9748,
+      "step": 17326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0131444968990096e-05,
+      "loss": 0.2678,
+      "step": 17328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0126816625011571e-05,
+      "loss": 7.7803,
+      "step": 17330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0122188281033047e-05,
+      "loss": 2.4892,
+      "step": 17332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0117559937054524e-05,
+      "loss": 0.0031,
+      "step": 17334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0112931593075999e-05,
+      "loss": 1.671,
+      "step": 17336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0108303249097473e-05,
+      "loss": 0.6398,
+      "step": 17338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0103674905118948e-05,
+      "loss": 5.6892,
+      "step": 17340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0099046561140424e-05,
+      "loss": 2.3067,
+      "step": 17342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0094418217161899e-05,
+      "loss": 1.477,
+      "step": 17344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0089789873183375e-05,
+      "loss": 0.121,
+      "step": 17346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.008516152920485e-05,
+      "loss": 2.1911,
+      "step": 17348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0080533185226327e-05,
+      "loss": 4.3986,
+      "step": 17350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0075904841247802e-05,
+      "loss": 1.3002,
+      "step": 17352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0071276497269278e-05,
+      "loss": 0.7485,
+      "step": 17354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0066648153290753e-05,
+      "loss": 2.5868,
+      "step": 17356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0062019809312229e-05,
+      "loss": 0.0467,
+      "step": 17358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0057391465333704e-05,
+      "loss": 0.5214,
+      "step": 17360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.005276312135518e-05,
+      "loss": 1.5878,
+      "step": 17362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0048134777376655e-05,
+      "loss": 0.1135,
+      "step": 17364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.004350643339813e-05,
+      "loss": 1.3467,
+      "step": 17366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0038878089419606e-05,
+      "loss": 3.5664,
+      "step": 17368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0034249745441083e-05,
+      "loss": 6.4193,
+      "step": 17370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0029621401462558e-05,
+      "loss": 3.404,
+      "step": 17372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0024993057484032e-05,
+      "loss": 4.2661,
+      "step": 17374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0020364713505508e-05,
+      "loss": 2.3746,
+      "step": 17376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0015736369526983e-05,
+      "loss": 0.6922,
+      "step": 17378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0011108025548458e-05,
+      "loss": 1.6739,
+      "step": 17380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.0006479681569934e-05,
+      "loss": 0.0015,
+      "step": 17382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.000185133759141e-05,
+      "loss": 0.7582,
+      "step": 17384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.997222993612885e-06,
+      "loss": 1.5105,
+      "step": 17386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.992594649634362e-06,
+      "loss": 2.7609,
+      "step": 17388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.987966305655837e-06,
+      "loss": 0.9346,
+      "step": 17390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.983337961677313e-06,
+      "loss": 1.5944,
+      "step": 17392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.978709617698788e-06,
+      "loss": 1.5516,
+      "step": 17394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.974081273720264e-06,
+      "loss": 1.1256,
+      "step": 17396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.969452929741739e-06,
+      "loss": 0.2813,
+      "step": 17398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.964824585763214e-06,
+      "loss": 3.4472,
+      "step": 17400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.96019624178469e-06,
+      "loss": 0.0025,
+      "step": 17402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.955567897806165e-06,
+      "loss": 0.6148,
+      "step": 17404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.95093955382764e-06,
+      "loss": 2.9507,
+      "step": 17406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.946311209849118e-06,
+      "loss": 0.0014,
+      "step": 17408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.941682865870593e-06,
+      "loss": 0.7553,
+      "step": 17410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.937054521892067e-06,
+      "loss": 1.3621,
+      "step": 17412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.932426177913542e-06,
+      "loss": 3.5511,
+      "step": 17414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.927797833935018e-06,
+      "loss": 0.007,
+      "step": 17416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.923169489956493e-06,
+      "loss": 3.5321,
+      "step": 17418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.918541145977969e-06,
+      "loss": 2.6825,
+      "step": 17420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.913912801999444e-06,
+      "loss": 0.4721,
+      "step": 17422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.909284458020921e-06,
+      "loss": 0.0037,
+      "step": 17424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.904656114042397e-06,
+      "loss": 1.0527,
+      "step": 17426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.900027770063872e-06,
+      "loss": 4.6482,
+      "step": 17428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.895399426085347e-06,
+      "loss": 0.0058,
+      "step": 17430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.890771082106823e-06,
+      "loss": 0.4187,
+      "step": 17432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.886142738128298e-06,
+      "loss": 0.0029,
+      "step": 17434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.881514394149774e-06,
+      "loss": 0.11,
+      "step": 17436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.876886050171249e-06,
+      "loss": 0.2489,
+      "step": 17438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.872257706192725e-06,
+      "loss": 2.8421,
+      "step": 17440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.8676293622142e-06,
+      "loss": 3.3258,
+      "step": 17442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.863001018235677e-06,
+      "loss": 0.0012,
+      "step": 17444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.858372674257153e-06,
+      "loss": 1.652,
+      "step": 17446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.853744330278626e-06,
+      "loss": 1.0631,
+      "step": 17448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.849115986300102e-06,
+      "loss": 1.1922,
+      "step": 17450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.844487642321577e-06,
+      "loss": 4.0656,
+      "step": 17452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.839859298343053e-06,
+      "loss": 4.4561,
+      "step": 17454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.835230954364528e-06,
+      "loss": 0.7948,
+      "step": 17456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.830602610386003e-06,
+      "loss": 4.165,
+      "step": 17458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.82597426640748e-06,
+      "loss": 0.2439,
+      "step": 17460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.821345922428956e-06,
+      "loss": 0.0006,
+      "step": 17462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.816717578450431e-06,
+      "loss": 1.4368,
+      "step": 17464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.812089234471907e-06,
+      "loss": 1.7057,
+      "step": 17466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.807460890493382e-06,
+      "loss": 4.28,
+      "step": 17468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.802832546514858e-06,
+      "loss": 0.0003,
+      "step": 17470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.798204202536333e-06,
+      "loss": 1.3202,
+      "step": 17472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.793575858557808e-06,
+      "loss": 1.9023,
+      "step": 17474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.788947514579284e-06,
+      "loss": 0.7103,
+      "step": 17476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.78431917060076e-06,
+      "loss": 6.2731,
+      "step": 17478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.779690826622236e-06,
+      "loss": 0.2228,
+      "step": 17480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.775062482643712e-06,
+      "loss": 2.0402,
+      "step": 17482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.770434138665186e-06,
+      "loss": 2.1915,
+      "step": 17484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.765805794686661e-06,
+      "loss": 0.2848,
+      "step": 17486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.761177450708136e-06,
+      "loss": 5.168,
+      "step": 17488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.756549106729612e-06,
+      "loss": 2.8916,
+      "step": 17490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.751920762751087e-06,
+      "loss": 3.355,
+      "step": 17492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.747292418772563e-06,
+      "loss": 0.1571,
+      "step": 17494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.742664074794038e-06,
+      "loss": 0.4299,
+      "step": 17496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.738035730815515e-06,
+      "loss": 1.1369,
+      "step": 17498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.73340738683699e-06,
+      "loss": 0.341,
+      "step": 17500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.728779042858466e-06,
+      "loss": 0.3027,
+      "step": 17502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.724150698879942e-06,
+      "loss": 0.0021,
+      "step": 17504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.719522354901417e-06,
+      "loss": 0.0442,
+      "step": 17506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.714894010922892e-06,
+      "loss": 4.8425,
+      "step": 17508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.710265666944368e-06,
+      "loss": 0.4896,
+      "step": 17510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.705637322965843e-06,
+      "loss": 3.6537,
+      "step": 17512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.701008978987319e-06,
+      "loss": 4.4468,
+      "step": 17514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.696380635008794e-06,
+      "loss": 0.04,
+      "step": 17516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.691752291030271e-06,
+      "loss": 1.3352,
+      "step": 17518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.687123947051745e-06,
+      "loss": 2.6653,
+      "step": 17520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.68249560307322e-06,
+      "loss": 0.1341,
+      "step": 17522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.677867259094696e-06,
+      "loss": 0.004,
+      "step": 17524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.673238915116171e-06,
+      "loss": 2.7277,
+      "step": 17526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.668610571137647e-06,
+      "loss": 0.8272,
+      "step": 17528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.663982227159122e-06,
+      "loss": 2.0371,
+      "step": 17530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.659353883180597e-06,
+      "loss": 0.5155,
+      "step": 17532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.654725539202075e-06,
+      "loss": 2.9593,
+      "step": 17534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.65009719522355e-06,
+      "loss": 4.2907,
+      "step": 17536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.645468851245025e-06,
+      "loss": 0.0016,
+      "step": 17538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.640840507266501e-06,
+      "loss": 5.2405,
+      "step": 17540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.636212163287976e-06,
+      "loss": 0.4696,
+      "step": 17542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.631583819309452e-06,
+      "loss": 0.0109,
+      "step": 17544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.626955475330927e-06,
+      "loss": 2.8925,
+      "step": 17546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.622327131352403e-06,
+      "loss": 0.9879,
+      "step": 17548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.617698787373878e-06,
+      "loss": 2.5846,
+      "step": 17550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.613070443395353e-06,
+      "loss": 2.135,
+      "step": 17552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.60844209941683e-06,
+      "loss": 4.5932,
+      "step": 17554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.603813755438304e-06,
+      "loss": 2.2842,
+      "step": 17556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.59918541145978e-06,
+      "loss": 0.0031,
+      "step": 17558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.594557067481255e-06,
+      "loss": 0.3737,
+      "step": 17560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.58992872350273e-06,
+      "loss": 3.5355,
+      "step": 17562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.585300379524206e-06,
+      "loss": 1.3505,
+      "step": 17564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.580672035545681e-06,
+      "loss": 2.0247,
+      "step": 17566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.576043691567157e-06,
+      "loss": 1.9294,
+      "step": 17568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.571415347588634e-06,
+      "loss": 0.6516,
+      "step": 17570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.56678700361011e-06,
+      "loss": 4.0319,
+      "step": 17572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.562158659631585e-06,
+      "loss": 1.9675,
+      "step": 17574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.55753031565306e-06,
+      "loss": 2.1134,
+      "step": 17576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.552901971674536e-06,
+      "loss": 6.9561,
+      "step": 17578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.548273627696011e-06,
+      "loss": 3.1191,
+      "step": 17580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.543645283717486e-06,
+      "loss": 0.6626,
+      "step": 17582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.539016939738962e-06,
+      "loss": 1.3087,
+      "step": 17584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.534388595760437e-06,
+      "loss": 2.3891,
+      "step": 17586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.529760251781913e-06,
+      "loss": 2.266,
+      "step": 17588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.52513190780339e-06,
+      "loss": 1.9666,
+      "step": 17590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.520503563824864e-06,
+      "loss": 0.0058,
+      "step": 17592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.515875219846339e-06,
+      "loss": 4.5148,
+      "step": 17594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.511246875867814e-06,
+      "loss": 0.2373,
+      "step": 17596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.50661853188929e-06,
+      "loss": 2.8274,
+      "step": 17598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.501990187910765e-06,
+      "loss": 2.6116,
+      "step": 17600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49736184393224e-06,
+      "loss": 2.3679,
+      "step": 17602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.492733499953716e-06,
+      "loss": 1.7609,
+      "step": 17604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.488105155975192e-06,
+      "loss": 0.6321,
+      "step": 17606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483476811996669e-06,
+      "loss": 0.0238,
+      "step": 17608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.478848468018144e-06,
+      "loss": 2.8545,
+      "step": 17610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47422012403962e-06,
+      "loss": 0.3889,
+      "step": 17612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.469591780061095e-06,
+      "loss": 5.1684,
+      "step": 17614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46496343608257e-06,
+      "loss": 0.9859,
+      "step": 17616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.460335092104046e-06,
+      "loss": 1.3481,
+      "step": 17618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.455706748125521e-06,
+      "loss": 2.5175,
+      "step": 17620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451078404146997e-06,
+      "loss": 0.0036,
+      "step": 17622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.446450060168472e-06,
+      "loss": 1.8033,
+      "step": 17624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.441821716189948e-06,
+      "loss": 2.4979,
+      "step": 17626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437193372211423e-06,
+      "loss": 0.0055,
+      "step": 17628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432565028232898e-06,
+      "loss": 0.017,
+      "step": 17630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.427936684254374e-06,
+      "loss": 0.6679,
+      "step": 17632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42330834027585e-06,
+      "loss": 1.5658,
+      "step": 17634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.418679996297325e-06,
+      "loss": 0.0829,
+      "step": 17636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4140516523188e-06,
+      "loss": 2.1522,
+      "step": 17638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.409423308340275e-06,
+      "loss": 1.8307,
+      "step": 17640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.404794964361751e-06,
+      "loss": 6.3148,
+      "step": 17642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.400166620383228e-06,
+      "loss": 3.6199,
+      "step": 17644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.395538276404703e-06,
+      "loss": 2.5513,
+      "step": 17646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.390909932426179e-06,
+      "loss": 0.5538,
+      "step": 17648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.386281588447654e-06,
+      "loss": 3.1203,
+      "step": 17650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38165324446913e-06,
+      "loss": 0.5367,
+      "step": 17652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.377024900490605e-06,
+      "loss": 0.6674,
+      "step": 17654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.37239655651208e-06,
+      "loss": 1.3472,
+      "step": 17656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.367768212533556e-06,
+      "loss": 4.3143,
+      "step": 17658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.363139868555031e-06,
+      "loss": 0.9241,
+      "step": 17660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.358511524576507e-06,
+      "loss": 0.876,
+      "step": 17662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.353883180597982e-06,
+      "loss": 1.5038,
+      "step": 17664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.349254836619458e-06,
+      "loss": 4.8952,
+      "step": 17666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.344626492640933e-06,
+      "loss": 1.2906,
+      "step": 17668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.339998148662409e-06,
+      "loss": 4.0824,
+      "step": 17670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.335369804683884e-06,
+      "loss": 2.5006,
+      "step": 17672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.33074146070536e-06,
+      "loss": 0.0025,
+      "step": 17674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.326113116726835e-06,
+      "loss": 1.7942,
+      "step": 17676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.32148477274831e-06,
+      "loss": 6.2133,
+      "step": 17678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.316856428769787e-06,
+      "loss": 0.0132,
+      "step": 17680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.312228084791263e-06,
+      "loss": 2.2574,
+      "step": 17682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.307599740812738e-06,
+      "loss": 0.6463,
+      "step": 17684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302971396834214e-06,
+      "loss": 1.0197,
+      "step": 17686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298343052855689e-06,
+      "loss": 1.8029,
+      "step": 17688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293714708877164e-06,
+      "loss": 0.0026,
+      "step": 17690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28908636489864e-06,
+      "loss": 4.6563,
+      "step": 17692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284458020920115e-06,
+      "loss": 1.4874,
+      "step": 17694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27982967694159e-06,
+      "loss": 1.423,
+      "step": 17696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275201332963066e-06,
+      "loss": 1.2733,
+      "step": 17698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270572988984542e-06,
+      "loss": 1.1392,
+      "step": 17700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265944645006017e-06,
+      "loss": 0.0025,
+      "step": 17702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261316301027492e-06,
+      "loss": 1.2373,
+      "step": 17704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256687957048968e-06,
+      "loss": 0.3966,
+      "step": 17706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252059613070443e-06,
+      "loss": 1.4258,
+      "step": 17708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247431269091919e-06,
+      "loss": 0.0012,
+      "step": 17710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242802925113394e-06,
+      "loss": 5.738,
+      "step": 17712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23817458113487e-06,
+      "loss": 0.5457,
+      "step": 17714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233546237156345e-06,
+      "loss": 0.0381,
+      "step": 17716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228917893177822e-06,
+      "loss": 1.7831,
+      "step": 17718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224289549199298e-06,
+      "loss": 0.2575,
+      "step": 17720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219661205220773e-06,
+      "loss": 0.0389,
+      "step": 17722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.215032861242248e-06,
+      "loss": 0.396,
+      "step": 17724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.210404517263724e-06,
+      "loss": 1.0754,
+      "step": 17726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2057761732852e-06,
+      "loss": 1.2545,
+      "step": 17728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.201147829306675e-06,
+      "loss": 1.8999,
+      "step": 17730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.19651948532815e-06,
+      "loss": 1.8618,
+      "step": 17732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.191891141349626e-06,
+      "loss": 1.5642,
+      "step": 17734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.187262797371101e-06,
+      "loss": 5.8555,
+      "step": 17736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.182634453392576e-06,
+      "loss": 1.414,
+      "step": 17738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.178006109414052e-06,
+      "loss": 0.835,
+      "step": 17740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.173377765435527e-06,
+      "loss": 0.0209,
+      "step": 17742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.168749421457003e-06,
+      "loss": 3.209,
+      "step": 17744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.164121077478478e-06,
+      "loss": 1.0643,
+      "step": 17746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.159492733499954e-06,
+      "loss": 0.0005,
+      "step": 17748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.154864389521429e-06,
+      "loss": 0.0025,
+      "step": 17750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.150236045542904e-06,
+      "loss": 0.019,
+      "step": 17752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.145607701564381e-06,
+      "loss": 1.0509,
+      "step": 17754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.140979357585857e-06,
+      "loss": 0.0287,
+      "step": 17756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.136351013607332e-06,
+      "loss": 2.5271,
+      "step": 17758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.131722669628808e-06,
+      "loss": 3.7968,
+      "step": 17760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.127094325650283e-06,
+      "loss": 0.0036,
+      "step": 17762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.122465981671759e-06,
+      "loss": 3.6761,
+      "step": 17764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.117837637693234e-06,
+      "loss": 1.4779,
+      "step": 17766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.11320929371471e-06,
+      "loss": 3.9365,
+      "step": 17768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.108580949736185e-06,
+      "loss": 0.0006,
+      "step": 17770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.10395260575766e-06,
+      "loss": 1.6952,
+      "step": 17772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.099324261779136e-06,
+      "loss": 0.0071,
+      "step": 17774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.094695917800611e-06,
+      "loss": 0.5095,
+      "step": 17776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090067573822087e-06,
+      "loss": 1.4908,
+      "step": 17778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.085439229843562e-06,
+      "loss": 0.0017,
+      "step": 17780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.080810885865037e-06,
+      "loss": 1.7061,
+      "step": 17782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.076182541886513e-06,
+      "loss": 0.5146,
+      "step": 17784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071554197907988e-06,
+      "loss": 0.2747,
+      "step": 17786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.066925853929464e-06,
+      "loss": 0.3484,
+      "step": 17788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.062297509950939e-06,
+      "loss": 0.0152,
+      "step": 17790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.057669165972416e-06,
+      "loss": 0.2273,
+      "step": 17792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.053040821993892e-06,
+      "loss": 4.9922,
+      "step": 17794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.048412478015367e-06,
+      "loss": 2.2244,
+      "step": 17796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.043784134036843e-06,
+      "loss": 1.0989,
+      "step": 17798
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.039155790058318e-06,
+      "loss": 2.4083,
+      "step": 17800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.034527446079793e-06,
+      "loss": 0.1054,
+      "step": 17802
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.029899102101269e-06,
+      "loss": 5.7005,
+      "step": 17804
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.025270758122744e-06,
+      "loss": 0.001,
+      "step": 17806
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.02064241414422e-06,
+      "loss": 0.279,
+      "step": 17808
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.016014070165695e-06,
+      "loss": 0.3224,
+      "step": 17810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01138572618717e-06,
+      "loss": 0.4758,
+      "step": 17812
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.006757382208646e-06,
+      "loss": 1.5108,
+      "step": 17814
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.002129038230121e-06,
+      "loss": 4.9447,
+      "step": 17816
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997500694251597e-06,
+      "loss": 0.9999,
+      "step": 17818
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992872350273072e-06,
+      "loss": 7.6745,
+      "step": 17820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.988244006294548e-06,
+      "loss": 0.0222,
+      "step": 17822
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.983615662316023e-06,
+      "loss": 0.2438,
+      "step": 17824
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978987318337498e-06,
+      "loss": 2.2164,
+      "step": 17826
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 6.5445,
+      "step": 17828
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.969730630380451e-06,
+      "loss": 0.097,
+      "step": 17830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.965102286401926e-06,
+      "loss": 1.0672,
+      "step": 17832
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.960473942423402e-06,
+      "loss": 2.4275,
+      "step": 17834
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.955845598444877e-06,
+      "loss": 1.7259,
+      "step": 17836
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.951217254466353e-06,
+      "loss": 2.8615,
+      "step": 17838
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.946588910487828e-06,
+      "loss": 4.5009,
+      "step": 17840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.941960566509304e-06,
+      "loss": 1.9145,
+      "step": 17842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.937332222530779e-06,
+      "loss": 2.1639,
+      "step": 17844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.932703878552254e-06,
+      "loss": 1.2489,
+      "step": 17846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.92807553457373e-06,
+      "loss": 0.0014,
+      "step": 17848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.923447190595205e-06,
+      "loss": 0.0007,
+      "step": 17850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.91881884661668e-06,
+      "loss": 1.8288,
+      "step": 17852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.914190502638156e-06,
+      "loss": 0.1261,
+      "step": 17854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.909562158659632e-06,
+      "loss": 3.5555,
+      "step": 17856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.904933814681107e-06,
+      "loss": 0.0037,
+      "step": 17858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.900305470702582e-06,
+      "loss": 0.877,
+      "step": 17860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.895677126724058e-06,
+      "loss": 1.5135,
+      "step": 17862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.891048782745535e-06,
+      "loss": 0.0045,
+      "step": 17864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.88642043876701e-06,
+      "loss": 0.0309,
+      "step": 17866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.881792094788486e-06,
+      "loss": 0.0038,
+      "step": 17868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.877163750809961e-06,
+      "loss": 2.268,
+      "step": 17870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.872535406831437e-06,
+      "loss": 0.8377,
+      "step": 17872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.867907062852912e-06,
+      "loss": 1.279,
+      "step": 17874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.863278718874387e-06,
+      "loss": 1.7032,
+      "step": 17876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.858650374895863e-06,
+      "loss": 0.2339,
+      "step": 17878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.854022030917338e-06,
+      "loss": 2.2193,
+      "step": 17880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.849393686938814e-06,
+      "loss": 0.011,
+      "step": 17882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.84476534296029e-06,
+      "loss": 1.4706,
+      "step": 17884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.840136998981765e-06,
+      "loss": 5.1827,
+      "step": 17886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.83550865500324e-06,
+      "loss": 3.4926,
+      "step": 17888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.830880311024715e-06,
+      "loss": 1.9671,
+      "step": 17890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.826251967046191e-06,
+      "loss": 0.775,
+      "step": 17892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.821623623067666e-06,
+      "loss": 3.3082,
+      "step": 17894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.816995279089142e-06,
+      "loss": 3.4971,
+      "step": 17896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.812366935110617e-06,
+      "loss": 0.929,
+      "step": 17898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.807738591132093e-06,
+      "loss": 5.519,
+      "step": 17900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.80311024715357e-06,
+      "loss": 0.8438,
+      "step": 17902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.798481903175045e-06,
+      "loss": 2.022,
+      "step": 17904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.79385355919652e-06,
+      "loss": 5.0338,
+      "step": 17906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.789225215217996e-06,
+      "loss": 3.1664,
+      "step": 17908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.784596871239471e-06,
+      "loss": 1.125,
+      "step": 17910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.779968527260947e-06,
+      "loss": 1.2995,
+      "step": 17912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.775340183282422e-06,
+      "loss": 0.7664,
+      "step": 17914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.770711839303898e-06,
+      "loss": 0.5544,
+      "step": 17916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.766083495325373e-06,
+      "loss": 2.2225,
+      "step": 17918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.761455151346848e-06,
+      "loss": 4.0363,
+      "step": 17920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.756826807368324e-06,
+      "loss": 1.248,
+      "step": 17922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.7521984633898e-06,
+      "loss": 4.646,
+      "step": 17924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.747570119411275e-06,
+      "loss": 0.0228,
+      "step": 17926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.74294177543275e-06,
+      "loss": 3.1551,
+      "step": 17928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.738313431454226e-06,
+      "loss": 0.5551,
+      "step": 17930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.733685087475701e-06,
+      "loss": 0.262,
+      "step": 17932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.729056743497176e-06,
+      "loss": 1.1281,
+      "step": 17934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.724428399518652e-06,
+      "loss": 0.727,
+      "step": 17936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.719800055540129e-06,
+      "loss": 1.2762,
+      "step": 17938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.715171711561604e-06,
+      "loss": 4.7664,
+      "step": 17940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.71054336758308e-06,
+      "loss": 3.4277,
+      "step": 17942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.705915023604555e-06,
+      "loss": 0.0556,
+      "step": 17944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.70128667962603e-06,
+      "loss": 1.2683,
+      "step": 17946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.696658335647506e-06,
+      "loss": 0.0596,
+      "step": 17948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.692029991668982e-06,
+      "loss": 1.8033,
+      "step": 17950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.687401647690457e-06,
+      "loss": 4.9111,
+      "step": 17952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.682773303711932e-06,
+      "loss": 0.6405,
+      "step": 17954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.678144959733408e-06,
+      "loss": 0.6897,
+      "step": 17956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.673516615754883e-06,
+      "loss": 0.333,
+      "step": 17958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.668888271776359e-06,
+      "loss": 1.2894,
+      "step": 17960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.664259927797834e-06,
+      "loss": 0.0086,
+      "step": 17962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.65963158381931e-06,
+      "loss": 0.0005,
+      "step": 17964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.655003239840785e-06,
+      "loss": 0.5376,
+      "step": 17966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.65037489586226e-06,
+      "loss": 1.3204,
+      "step": 17968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.645746551883736e-06,
+      "loss": 3.2096,
+      "step": 17970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.641118207905211e-06,
+      "loss": 0.725,
+      "step": 17972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.636489863926688e-06,
+      "loss": 0.2159,
+      "step": 17974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.631861519948164e-06,
+      "loss": 7.5116,
+      "step": 17976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.62723317596964e-06,
+      "loss": 1.0297,
+      "step": 17978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.622604831991115e-06,
+      "loss": 0.2622,
+      "step": 17980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.61797648801259e-06,
+      "loss": 0.0005,
+      "step": 17982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.613348144034065e-06,
+      "loss": 2.2767,
+      "step": 17984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.608719800055541e-06,
+      "loss": 0.4739,
+      "step": 17986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.604091456077016e-06,
+      "loss": 0.3689,
+      "step": 17988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.599463112098492e-06,
+      "loss": 2.0531,
+      "step": 17990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.594834768119967e-06,
+      "loss": 0.0147,
+      "step": 17992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.590206424141443e-06,
+      "loss": 2.032,
+      "step": 17994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.585578080162918e-06,
+      "loss": 1.4323,
+      "step": 17996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.580949736184393e-06,
+      "loss": 0.4126,
+      "step": 17998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.576321392205869e-06,
+      "loss": 0.0003,
+      "step": 18000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.571693048227344e-06,
+      "loss": 0.1226,
+      "step": 18002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.56706470424882e-06,
+      "loss": 0.0009,
+      "step": 18004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.562436360270295e-06,
+      "loss": 0.2063,
+      "step": 18006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.55780801629177e-06,
+      "loss": 3.1553,
+      "step": 18008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.553179672313246e-06,
+      "loss": 3.64,
+      "step": 18010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.548551328334723e-06,
+      "loss": 0.0023,
+      "step": 18012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.543922984356199e-06,
+      "loss": 0.1195,
+      "step": 18014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.539294640377674e-06,
+      "loss": 1.8913,
+      "step": 18016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.53466629639915e-06,
+      "loss": 2.9787,
+      "step": 18018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.530037952420625e-06,
+      "loss": 8.5651,
+      "step": 18020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.5254096084421e-06,
+      "loss": 0.2374,
+      "step": 18022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.520781264463576e-06,
+      "loss": 2.8886,
+      "step": 18024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.516152920485051e-06,
+      "loss": 1.0926,
+      "step": 18026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.511524576506526e-06,
+      "loss": 1.2612,
+      "step": 18028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.506896232528002e-06,
+      "loss": 0.7364,
+      "step": 18030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.502267888549477e-06,
+      "loss": 0.2961,
+      "step": 18032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.497639544570953e-06,
+      "loss": 0.6282,
+      "step": 18034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.493011200592428e-06,
+      "loss": 1.8184,
+      "step": 18036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.488382856613904e-06,
+      "loss": 0.7538,
+      "step": 18038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.483754512635379e-06,
+      "loss": 0.0014,
+      "step": 18040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.479126168656854e-06,
+      "loss": 3.8086,
+      "step": 18042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.47449782467833e-06,
+      "loss": 0.097,
+      "step": 18044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.469869480699805e-06,
+      "loss": 1.3109,
+      "step": 18046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.465241136721282e-06,
+      "loss": 0.2929,
+      "step": 18048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.460612792742758e-06,
+      "loss": 2.4523,
+      "step": 18050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.455984448764233e-06,
+      "loss": 5.3084,
+      "step": 18052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.451356104785709e-06,
+      "loss": 4.6654,
+      "step": 18054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.446727760807184e-06,
+      "loss": 0.4405,
+      "step": 18056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.44209941682866e-06,
+      "loss": 0.0006,
+      "step": 18058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.437471072850135e-06,
+      "loss": 5.9472,
+      "step": 18060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.43284272887161e-06,
+      "loss": 4.2324,
+      "step": 18062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.428214384893086e-06,
+      "loss": 2.5635,
+      "step": 18064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.423586040914561e-06,
+      "loss": 0.1121,
+      "step": 18066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.418957696936037e-06,
+      "loss": 1.9448,
+      "step": 18068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.414329352957512e-06,
+      "loss": 0.0143,
+      "step": 18070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.409701008978988e-06,
+      "loss": 0.0027,
+      "step": 18072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.405072665000463e-06,
+      "loss": 0.0016,
+      "step": 18074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.400444321021938e-06,
+      "loss": 0.1151,
+      "step": 18076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.395815977043414e-06,
+      "loss": 1.4308,
+      "step": 18078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.39118763306489e-06,
+      "loss": 5.6127,
+      "step": 18080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.386559289086365e-06,
+      "loss": 1.9714,
+      "step": 18082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.381930945107842e-06,
+      "loss": 0.027,
+      "step": 18084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.377302601129317e-06,
+      "loss": 1.4655,
+      "step": 18086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.372674257150793e-06,
+      "loss": 2.5395,
+      "step": 18088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.368045913172268e-06,
+      "loss": 6.8316,
+      "step": 18090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.363417569193743e-06,
+      "loss": 0.8381,
+      "step": 18092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.358789225215219e-06,
+      "loss": 2.0222,
+      "step": 18094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.354160881236694e-06,
+      "loss": 3.5211,
+      "step": 18096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.34953253725817e-06,
+      "loss": 0.5721,
+      "step": 18098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.344904193279643e-06,
+      "loss": 1.6093,
+      "step": 18100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.34027584930112e-06,
+      "loss": 2.403,
+      "step": 18102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.335647505322596e-06,
+      "loss": 0.9236,
+      "step": 18104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.331019161344071e-06,
+      "loss": 0.0016,
+      "step": 18106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.326390817365547e-06,
+      "loss": 1.4603,
+      "step": 18108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.321762473387022e-06,
+      "loss": 1.7034,
+      "step": 18110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.317134129408498e-06,
+      "loss": 1.9037,
+      "step": 18112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.312505785429973e-06,
+      "loss": 0.3992,
+      "step": 18114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.307877441451449e-06,
+      "loss": 1.3445,
+      "step": 18116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.303249097472924e-06,
+      "loss": 1.2552,
+      "step": 18118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.2986207534944e-06,
+      "loss": 0.0005,
+      "step": 18120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.293992409515877e-06,
+      "loss": 1.7361,
+      "step": 18122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.289364065537352e-06,
+      "loss": 0.6006,
+      "step": 18124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.284735721558827e-06,
+      "loss": 0.5287,
+      "step": 18126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.280107377580303e-06,
+      "loss": 0.1347,
+      "step": 18128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.275479033601778e-06,
+      "loss": 1.2993,
+      "step": 18130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.270850689623254e-06,
+      "loss": 0.0012,
+      "step": 18132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.266222345644729e-06,
+      "loss": 2.9028,
+      "step": 18134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.261594001666203e-06,
+      "loss": 4.2111,
+      "step": 18136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.25696565768768e-06,
+      "loss": 1.9959,
+      "step": 18138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.252337313709155e-06,
+      "loss": 0.7133,
+      "step": 18140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.24770896973063e-06,
+      "loss": 0.0013,
+      "step": 18142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.243080625752106e-06,
+      "loss": 0.7486,
+      "step": 18144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.238452281773582e-06,
+      "loss": 0.3887,
+      "step": 18146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.233823937795057e-06,
+      "loss": 0.9095,
+      "step": 18148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.229195593816532e-06,
+      "loss": 1.7729,
+      "step": 18150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.224567249838008e-06,
+      "loss": 0.5791,
+      "step": 18152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.219938905859483e-06,
+      "loss": 4.6596,
+      "step": 18154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.215310561880959e-06,
+      "loss": 0.0357,
+      "step": 18156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.210682217902436e-06,
+      "loss": 1.7598,
+      "step": 18158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.206053873923911e-06,
+      "loss": 2.1792,
+      "step": 18160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.201425529945387e-06,
+      "loss": 0.0645,
+      "step": 18162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.196797185966862e-06,
+      "loss": 0.0007,
+      "step": 18164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.192168841988338e-06,
+      "loss": 2.0623,
+      "step": 18166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.187540498009813e-06,
+      "loss": 7.3585,
+      "step": 18168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.182912154031288e-06,
+      "loss": 0.0008,
+      "step": 18170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.178283810052762e-06,
+      "loss": 4.7933,
+      "step": 18172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.17365546607424e-06,
+      "loss": 1.4122,
+      "step": 18174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.169027122095715e-06,
+      "loss": 7.7284,
+      "step": 18176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.16439877811719e-06,
+      "loss": 1.1591,
+      "step": 18178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.159770434138666e-06,
+      "loss": 2.8168,
+      "step": 18180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.155142090160141e-06,
+      "loss": 0.5556,
+      "step": 18182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.150513746181616e-06,
+      "loss": 2.487,
+      "step": 18184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.145885402203092e-06,
+      "loss": 0.2101,
+      "step": 18186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.141257058224567e-06,
+      "loss": 0.6281,
+      "step": 18188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.136628714246043e-06,
+      "loss": 2.1692,
+      "step": 18190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.132000370267518e-06,
+      "loss": 4.0518,
+      "step": 18192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.127372026288994e-06,
+      "loss": 1.1794,
+      "step": 18194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.12274368231047e-06,
+      "loss": 0.0047,
+      "step": 18196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.118115338331946e-06,
+      "loss": 3.7098,
+      "step": 18198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.113486994353421e-06,
+      "loss": 0.2851,
+      "step": 18200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.108858650374897e-06,
+      "loss": 0.8921,
+      "step": 18202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.104230306396372e-06,
+      "loss": 0.0077,
+      "step": 18204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.099601962417848e-06,
+      "loss": 2.8095,
+      "step": 18206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.094973618439321e-06,
+      "loss": 2.9735,
+      "step": 18208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.090345274460797e-06,
+      "loss": 2.5299,
+      "step": 18210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.085716930482274e-06,
+      "loss": 0.0008,
+      "step": 18212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.08108858650375e-06,
+      "loss": 1.5211,
+      "step": 18214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.076460242525225e-06,
+      "loss": 2.054,
+      "step": 18216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.0718318985467e-06,
+      "loss": 0.2518,
+      "step": 18218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.067203554568176e-06,
+      "loss": 1.696,
+      "step": 18220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.062575210589651e-06,
+      "loss": 0.2729,
+      "step": 18222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.057946866611127e-06,
+      "loss": 2.0137,
+      "step": 18224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.053318522632602e-06,
+      "loss": 1.4195,
+      "step": 18226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.048690178654077e-06,
+      "loss": 1.1337,
+      "step": 18228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.044061834675553e-06,
+      "loss": 3.2558,
+      "step": 18230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.03943349069703e-06,
+      "loss": 1.7845,
+      "step": 18232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.034805146718505e-06,
+      "loss": 4.2132,
+      "step": 18234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.03017680273998e-06,
+      "loss": 1.1297,
+      "step": 18236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.025548458761456e-06,
+      "loss": 1.6473,
+      "step": 18238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.020920114782932e-06,
+      "loss": 4.6807,
+      "step": 18240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.016291770804407e-06,
+      "loss": 4.8673,
+      "step": 18242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.01166342682588e-06,
+      "loss": 2.3331,
+      "step": 18244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.007035082847356e-06,
+      "loss": 0.0358,
+      "step": 18246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.002406738868833e-06,
+      "loss": 0.0444,
+      "step": 18248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.997778394890309e-06,
+      "loss": 0.6343,
+      "step": 18250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.993150050911784e-06,
+      "loss": 2.3068,
+      "step": 18252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.98852170693326e-06,
+      "loss": 2.1088,
+      "step": 18254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.983893362954735e-06,
+      "loss": 4.7567,
+      "step": 18256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.97926501897621e-06,
+      "loss": 1.5881,
+      "step": 18258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.974636674997686e-06,
+      "loss": 0.4313,
+      "step": 18260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.970008331019161e-06,
+      "loss": 0.8834,
+      "step": 18262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.965379987040637e-06,
+      "loss": 1.7597,
+      "step": 18264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.960751643062112e-06,
+      "loss": 0.8536,
+      "step": 18266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.95612329908359e-06,
+      "loss": 0.0007,
+      "step": 18268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.951494955105065e-06,
+      "loss": 1.5067,
+      "step": 18270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.94686661112654e-06,
+      "loss": 1.1456,
+      "step": 18272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.942238267148016e-06,
+      "loss": 1.4295,
+      "step": 18274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.937609923169491e-06,
+      "loss": 0.2277,
+      "step": 18276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.932981579190966e-06,
+      "loss": 1.969,
+      "step": 18278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.928353235212442e-06,
+      "loss": 0.2174,
+      "step": 18280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.923724891233916e-06,
+      "loss": 1.5742,
+      "step": 18282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.919096547255391e-06,
+      "loss": 3.232,
+      "step": 18284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.914468203276868e-06,
+      "loss": 7.8422,
+      "step": 18286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.909839859298344e-06,
+      "loss": 0.007,
+      "step": 18288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.905211515319819e-06,
+      "loss": 0.1761,
+      "step": 18290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.900583171341294e-06,
+      "loss": 1.4063,
+      "step": 18292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.89595482736277e-06,
+      "loss": 0.2543,
+      "step": 18294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.891326483384245e-06,
+      "loss": 0.1522,
+      "step": 18296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.88669813940572e-06,
+      "loss": 0.4265,
+      "step": 18298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.882069795427196e-06,
+      "loss": 0.6211,
+      "step": 18300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.877441451448672e-06,
+      "loss": 3.023,
+      "step": 18302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.872813107470147e-06,
+      "loss": 2.8604,
+      "step": 18304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.868184763491624e-06,
+      "loss": 0.1831,
+      "step": 18306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.8635564195131e-06,
+      "loss": 1.0248,
+      "step": 18308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.858928075534575e-06,
+      "loss": 2.515,
+      "step": 18310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.85429973155605e-06,
+      "loss": 2.4721,
+      "step": 18312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.849671387577526e-06,
+      "loss": 1.0742,
+      "step": 18314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.845043043599001e-06,
+      "loss": 0.9671,
+      "step": 18316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.840414699620475e-06,
+      "loss": 0.0049,
+      "step": 18318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.83578635564195e-06,
+      "loss": 2.3331,
+      "step": 18320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.831158011663427e-06,
+      "loss": 2.4793,
+      "step": 18322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.826529667684903e-06,
+      "loss": 2.3554,
+      "step": 18324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.821901323706378e-06,
+      "loss": 1.4372,
+      "step": 18326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.817272979727854e-06,
+      "loss": 2.2361,
+      "step": 18328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.81264463574933e-06,
+      "loss": 4.4393,
+      "step": 18330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.808016291770805e-06,
+      "loss": 4.7919,
+      "step": 18332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.80338794779228e-06,
+      "loss": 1.6699,
+      "step": 18334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.798759603813755e-06,
+      "loss": 0.0027,
+      "step": 18336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.794131259835231e-06,
+      "loss": 0.7938,
+      "step": 18338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.789502915856706e-06,
+      "loss": 1.9922,
+      "step": 18340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.784874571878183e-06,
+      "loss": 0.63,
+      "step": 18342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.780246227899659e-06,
+      "loss": 1.0586,
+      "step": 18344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.775617883921134e-06,
+      "loss": 1.5097,
+      "step": 18346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.77098953994261e-06,
+      "loss": 1.1844,
+      "step": 18348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.766361195964085e-06,
+      "loss": 1.1932,
+      "step": 18350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.76173285198556e-06,
+      "loss": 0.0024,
+      "step": 18352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.757104508007034e-06,
+      "loss": 1.8673,
+      "step": 18354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.75247616402851e-06,
+      "loss": 0.8251,
+      "step": 18356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.747847820049987e-06,
+      "loss": 3.9533,
+      "step": 18358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.743219476071462e-06,
+      "loss": 3.8987,
+      "step": 18360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.738591132092938e-06,
+      "loss": 1.2861,
+      "step": 18362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.733962788114413e-06,
+      "loss": 3.0504,
+      "step": 18364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.729334444135888e-06,
+      "loss": 2.6219,
+      "step": 18366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.724706100157364e-06,
+      "loss": 0.0261,
+      "step": 18368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.72007775617884e-06,
+      "loss": 0.0004,
+      "step": 18370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.715449412200315e-06,
+      "loss": 0.4662,
+      "step": 18372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.71082106822179e-06,
+      "loss": 0.0007,
+      "step": 18374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.706192724243266e-06,
+      "loss": 1.1284,
+      "step": 18376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.701564380264743e-06,
+      "loss": 0.3812,
+      "step": 18378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.696936036286218e-06,
+      "loss": 0.0439,
+      "step": 18380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.5613,
+      "step": 18382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.687679348329169e-06,
+      "loss": 1.1631,
+      "step": 18384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.683051004350644e-06,
+      "loss": 0.0007,
+      "step": 18386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.67842266037212e-06,
+      "loss": 3.1827,
+      "step": 18388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.673794316393594e-06,
+      "loss": 1.9194,
+      "step": 18390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.669165972415069e-06,
+      "loss": 0.037,
+      "step": 18392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.664537628436544e-06,
+      "loss": 1.0811,
+      "step": 18394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.659909284458022e-06,
+      "loss": 3.1579,
+      "step": 18396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.655280940479497e-06,
+      "loss": 1.4598,
+      "step": 18398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.650652596500972e-06,
+      "loss": 0.4743,
+      "step": 18400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.646024252522448e-06,
+      "loss": 3.2844,
+      "step": 18402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.641395908543923e-06,
+      "loss": 0.3954,
+      "step": 18404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.636767564565399e-06,
+      "loss": 0.4711,
+      "step": 18406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.632139220586874e-06,
+      "loss": 0.0012,
+      "step": 18408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.62751087660835e-06,
+      "loss": 1.853,
+      "step": 18410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.622882532629826e-06,
+      "loss": 0.0004,
+      "step": 18412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.618254188651301e-06,
+      "loss": 0.0032,
+      "step": 18414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.613625844672777e-06,
+      "loss": 5.5709,
+      "step": 18416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.608997500694252e-06,
+      "loss": 1.3824,
+      "step": 18418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.604369156715728e-06,
+      "loss": 4.7715,
+      "step": 18420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.599740812737204e-06,
+      "loss": 0.0002,
+      "step": 18422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.595112468758679e-06,
+      "loss": 0.0005,
+      "step": 18424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.590484124780154e-06,
+      "loss": 0.3506,
+      "step": 18426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.585855780801629e-06,
+      "loss": 2.0523,
+      "step": 18428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.581227436823105e-06,
+      "loss": 4.0864,
+      "step": 18430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.57659909284458e-06,
+      "loss": 4.7475,
+      "step": 18432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.5719707488660555e-06,
+      "loss": 0.3653,
+      "step": 18434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.567342404887532e-06,
+      "loss": 0.0459,
+      "step": 18436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.562714060909007e-06,
+      "loss": 1.9795,
+      "step": 18438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.558085716930483e-06,
+      "loss": 0.0398,
+      "step": 18440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.553457372951958e-06,
+      "loss": 2.1417,
+      "step": 18442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.5488290289734334e-06,
+      "loss": 1.1707,
+      "step": 18444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.544200684994909e-06,
+      "loss": 0.1611,
+      "step": 18446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.539572341016385e-06,
+      "loss": 0.0261,
+      "step": 18448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.5349439970378606e-06,
+      "loss": 0.0019,
+      "step": 18450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.530315653059336e-06,
+      "loss": 0.0013,
+      "step": 18452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.525687309080811e-06,
+      "loss": 2.0685,
+      "step": 18454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.521058965102287e-06,
+      "loss": 3.724,
+      "step": 18456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.516430621123763e-06,
+      "loss": 0.8842,
+      "step": 18458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.5118022771452385e-06,
+      "loss": 0.0009,
+      "step": 18460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.507173933166712e-06,
+      "loss": 1.9587,
+      "step": 18462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.5025455891881885e-06,
+      "loss": 1.3654,
+      "step": 18464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.497917245209664e-06,
+      "loss": 2.0614,
+      "step": 18466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.493288901231139e-06,
+      "loss": 0.0033,
+      "step": 18468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.488660557252615e-06,
+      "loss": 0.013,
+      "step": 18470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.48403221327409e-06,
+      "loss": 2.2166,
+      "step": 18472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4794038692955665e-06,
+      "loss": 2.2783,
+      "step": 18474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.474775525317042e-06,
+      "loss": 0.3097,
+      "step": 18476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.470147181338517e-06,
+      "loss": 0.158,
+      "step": 18478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.465518837359993e-06,
+      "loss": 0.0015,
+      "step": 18480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.460890493381468e-06,
+      "loss": 2.4144,
+      "step": 18482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4562621494029445e-06,
+      "loss": 6.3433,
+      "step": 18484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.45163380542442e-06,
+      "loss": 6.8526,
+      "step": 18486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.447005461445895e-06,
+      "loss": 2.0841,
+      "step": 18488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.442377117467371e-06,
+      "loss": 0.9099,
+      "step": 18490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.437748773488846e-06,
+      "loss": 1.4458,
+      "step": 18492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4331204295103224e-06,
+      "loss": 0.0007,
+      "step": 18494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.428492085531798e-06,
+      "loss": 1.7042,
+      "step": 18496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.423863741553272e-06,
+      "loss": 0.9089,
+      "step": 18498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.419235397574748e-06,
+      "loss": 2.7362,
+      "step": 18500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.414607053596223e-06,
+      "loss": 1.2885,
+      "step": 18502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.409978709617699e-06,
+      "loss": 0.0008,
+      "step": 18504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.405350365639174e-06,
+      "loss": 0.0435,
+      "step": 18506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4007220216606496e-06,
+      "loss": 2.9914,
+      "step": 18508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.396093677682126e-06,
+      "loss": 1.4342,
+      "step": 18510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.391465333703601e-06,
+      "loss": 0.6099,
+      "step": 18512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.386836989725077e-06,
+      "loss": 0.0023,
+      "step": 18514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.382208645746552e-06,
+      "loss": 5.5307,
+      "step": 18516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3775803017680275e-06,
+      "loss": 1.8235,
+      "step": 18518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.372951957789504e-06,
+      "loss": 0.7563,
+      "step": 18520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.368323613810979e-06,
+      "loss": 0.0007,
+      "step": 18522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.363695269832455e-06,
+      "loss": 2.7501,
+      "step": 18524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.35906692585393e-06,
+      "loss": 1.3935,
+      "step": 18526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3544385818754055e-06,
+      "loss": 3.2219,
+      "step": 18528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.349810237896882e-06,
+      "loss": 2.9072,
+      "step": 18530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.345181893918357e-06,
+      "loss": 0.7848,
+      "step": 18532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.340553549939831e-06,
+      "loss": 2.0571,
+      "step": 18534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.335925205961306e-06,
+      "loss": 1.5454,
+      "step": 18536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.331296861982783e-06,
+      "loss": 1.5176,
+      "step": 18538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.326668518004258e-06,
+      "loss": 2.8585,
+      "step": 18540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3220401740257335e-06,
+      "loss": 2.1632,
+      "step": 18542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.317411830047209e-06,
+      "loss": 3.3527,
+      "step": 18544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.312783486068684e-06,
+      "loss": 1.9906,
+      "step": 18546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.308155142090161e-06,
+      "loss": 0.2801,
+      "step": 18548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.303526798111636e-06,
+      "loss": 0.4718,
+      "step": 18550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.2988984541331115e-06,
+      "loss": 1.5365,
+      "step": 18552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.294270110154587e-06,
+      "loss": 4.0193,
+      "step": 18554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.289641766176062e-06,
+      "loss": 0.0081,
+      "step": 18556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.2850134221975386e-06,
+      "loss": 0.9081,
+      "step": 18558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.280385078219014e-06,
+      "loss": 2.3506,
+      "step": 18560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.275756734240489e-06,
+      "loss": 1.4049,
+      "step": 18562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.271128390261965e-06,
+      "loss": 0.0064,
+      "step": 18564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.26650004628344e-06,
+      "loss": 0.4414,
+      "step": 18566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.2618717023049165e-06,
+      "loss": 0.0008,
+      "step": 18568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.257243358326392e-06,
+      "loss": 3.2717,
+      "step": 18570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.252615014347866e-06,
+      "loss": 3.3214,
+      "step": 18572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.247986670369342e-06,
+      "loss": 1.9879,
+      "step": 18574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.243358326390817e-06,
+      "loss": 0.0429,
+      "step": 18576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.238729982412293e-06,
+      "loss": 1.3841,
+      "step": 18578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.234101638433768e-06,
+      "loss": 3.7469,
+      "step": 18580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.229473294455244e-06,
+      "loss": 3.26,
+      "step": 18582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.22484495047672e-06,
+      "loss": 0.0007,
+      "step": 18584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.220216606498195e-06,
+      "loss": 0.0692,
+      "step": 18586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.215588262519671e-06,
+      "loss": 0.8172,
+      "step": 18588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.210959918541146e-06,
+      "loss": 4.3344,
+      "step": 18590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.206331574562622e-06,
+      "loss": 0.2456,
+      "step": 18592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.201703230584098e-06,
+      "loss": 1.149,
+      "step": 18594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.197074886605573e-06,
+      "loss": 0.0015,
+      "step": 18596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.192446542627049e-06,
+      "loss": 3.3058,
+      "step": 18598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.187818198648524e-06,
+      "loss": 0.3084,
+      "step": 18600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.18318985467e-06,
+      "loss": 3.6076,
+      "step": 18602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.178561510691476e-06,
+      "loss": 0.5197,
+      "step": 18604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.173933166712951e-06,
+      "loss": 1.9619,
+      "step": 18606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.169304822734425e-06,
+      "loss": 1.9337,
+      "step": 18608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.164676478755901e-06,
+      "loss": 4.9472,
+      "step": 18610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.160048134777377e-06,
+      "loss": 3.9692,
+      "step": 18612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.155419790798852e-06,
+      "loss": 0.3928,
+      "step": 18614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.150791446820328e-06,
+      "loss": 0.0595,
+      "step": 18616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.146163102841803e-06,
+      "loss": 1.2091,
+      "step": 18618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.141534758863279e-06,
+      "loss": 0.0041,
+      "step": 18620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.136906414884755e-06,
+      "loss": 4.8461,
+      "step": 18622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.13227807090623e-06,
+      "loss": 0.1339,
+      "step": 18624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1276497269277056e-06,
+      "loss": 6.9697,
+      "step": 18626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.123021382949181e-06,
+      "loss": 0.0042,
+      "step": 18628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.118393038970657e-06,
+      "loss": 0.4483,
+      "step": 18630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.113764694992133e-06,
+      "loss": 0.0177,
+      "step": 18632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.109136351013608e-06,
+      "loss": 0.0901,
+      "step": 18634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1045080070350835e-06,
+      "loss": 0.7892,
+      "step": 18636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.099879663056559e-06,
+      "loss": 4.5744,
+      "step": 18638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.095251319078035e-06,
+      "loss": 2.1296,
+      "step": 18640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.090622975099511e-06,
+      "loss": 0.4795,
+      "step": 18642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.085994631120984e-06,
+      "loss": 3.8105,
+      "step": 18644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.08136628714246e-06,
+      "loss": 0.0017,
+      "step": 18646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.076737943163936e-06,
+      "loss": 1.3816,
+      "step": 18648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0721095991854115e-06,
+      "loss": 7.4232,
+      "step": 18650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.067481255206887e-06,
+      "loss": 1.1092,
+      "step": 18652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.062852911228362e-06,
+      "loss": 0.7366,
+      "step": 18654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.058224567249838e-06,
+      "loss": 1.5642,
+      "step": 18656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.053596223271314e-06,
+      "loss": 0.0066,
+      "step": 18658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0489678792927895e-06,
+      "loss": 0.0004,
+      "step": 18660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.044339535314265e-06,
+      "loss": 0.0005,
+      "step": 18662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.03971119133574e-06,
+      "loss": 1.8117,
+      "step": 18664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.035082847357216e-06,
+      "loss": 0.0208,
+      "step": 18666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.030454503378692e-06,
+      "loss": 0.3306,
+      "step": 18668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.0258261594001674e-06,
+      "loss": 0.5434,
+      "step": 18670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.021197815421643e-06,
+      "loss": 0.4323,
+      "step": 18672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.016569471443118e-06,
+      "loss": 0.0061,
+      "step": 18674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.011941127464594e-06,
+      "loss": 0.3397,
+      "step": 18676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.00731278348607e-06,
+      "loss": 0.524,
+      "step": 18678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.002684439507544e-06,
+      "loss": 5.4875,
+      "step": 18680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.998056095529019e-06,
+      "loss": 1.3183,
+      "step": 18682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.993427751550495e-06,
+      "loss": 0.1394,
+      "step": 18684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.988799407571971e-06,
+      "loss": 2.4119,
+      "step": 18686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.984171063593446e-06,
+      "loss": 1.3695,
+      "step": 18688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.979542719614922e-06,
+      "loss": 0.0846,
+      "step": 18690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.974914375636397e-06,
+      "loss": 3.8513,
+      "step": 18692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.970286031657873e-06,
+      "loss": 2.5898,
+      "step": 18694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.965657687679349e-06,
+      "loss": 0.0653,
+      "step": 18696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.961029343700824e-06,
+      "loss": 0.9791,
+      "step": 18698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.9564009997223e-06,
+      "loss": 0.0056,
+      "step": 18700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.951772655743775e-06,
+      "loss": 0.9113,
+      "step": 18702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.947144311765251e-06,
+      "loss": 0.1173,
+      "step": 18704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.942515967786727e-06,
+      "loss": 1.4396,
+      "step": 18706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.937887623808202e-06,
+      "loss": 1.5741,
+      "step": 18708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.933259279829678e-06,
+      "loss": 0.0006,
+      "step": 18710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.928630935851153e-06,
+      "loss": 1.6112,
+      "step": 18712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.924002591872629e-06,
+      "loss": 0.1142,
+      "step": 18714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.919374247894103e-06,
+      "loss": 1.2048,
+      "step": 18716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.9147459039155785e-06,
+      "loss": 0.2917,
+      "step": 18718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.910117559937055e-06,
+      "loss": 0.0033,
+      "step": 18720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.90548921595853e-06,
+      "loss": 0.3018,
+      "step": 18722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.900860871980006e-06,
+      "loss": 0.0035,
+      "step": 18724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.896232528001481e-06,
+      "loss": 2.7534,
+      "step": 18726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.8916041840229564e-06,
+      "loss": 3.4132,
+      "step": 18728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.886975840044433e-06,
+      "loss": 3.0728,
+      "step": 18730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.882347496065908e-06,
+      "loss": 6.0762,
+      "step": 18732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.8777191520873836e-06,
+      "loss": 3.4295,
+      "step": 18734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.873090808108859e-06,
+      "loss": 0.0007,
+      "step": 18736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.868462464130334e-06,
+      "loss": 2.202,
+      "step": 18738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.863834120151811e-06,
+      "loss": 1.2101,
+      "step": 18740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.859205776173286e-06,
+      "loss": 1.9056,
+      "step": 18742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.8545774321947615e-06,
+      "loss": 3.2887,
+      "step": 18744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.849949088216237e-06,
+      "loss": 0.0013,
+      "step": 18746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.845320744237712e-06,
+      "loss": 0.0007,
+      "step": 18748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.840692400259189e-06,
+      "loss": 1.0945,
+      "step": 18750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.836064056280662e-06,
+      "loss": 3.9319,
+      "step": 18752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.831435712302138e-06,
+      "loss": 0.388,
+      "step": 18754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.826807368323613e-06,
+      "loss": 2.751,
+      "step": 18756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.8221790243450895e-06,
+      "loss": 3.6404,
+      "step": 18758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.817550680366565e-06,
+      "loss": 3.1234,
+      "step": 18760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.81292233638804e-06,
+      "loss": 0.1499,
+      "step": 18762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.808293992409516e-06,
+      "loss": 2.6571,
+      "step": 18764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.803665648430991e-06,
+      "loss": 0.0037,
+      "step": 18766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.7990373044524675e-06,
+      "loss": 1.5199,
+      "step": 18768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.794408960473943e-06,
+      "loss": 1.3351,
+      "step": 18770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.789780616495418e-06,
+      "loss": 0.0045,
+      "step": 18772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.785152272516894e-06,
+      "loss": 5.2442,
+      "step": 18774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.780523928538369e-06,
+      "loss": 3.7629,
+      "step": 18776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.7758955845598454e-06,
+      "loss": 1.7444,
+      "step": 18778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.771267240581321e-06,
+      "loss": 1.83,
+      "step": 18780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.766638896602796e-06,
+      "loss": 1.7647,
+      "step": 18782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.762010552624272e-06,
+      "loss": 1.1358,
+      "step": 18784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.757382208645747e-06,
+      "loss": 1.987,
+      "step": 18786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.752753864667222e-06,
+      "loss": 4.1836,
+      "step": 18788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.748125520688697e-06,
+      "loss": 0.3735,
+      "step": 18790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.7434971767101726e-06,
+      "loss": 0.0028,
+      "step": 18792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.738868832731649e-06,
+      "loss": 5.9772,
+      "step": 18794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.734240488753124e-06,
+      "loss": 1.2038,
+      "step": 18796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.7296121447746e-06,
+      "loss": 1.7382,
+      "step": 18798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.724983800796075e-06,
+      "loss": 0.5711,
+      "step": 18800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.7203554568175505e-06,
+      "loss": 0.0007,
+      "step": 18802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.715727112839027e-06,
+      "loss": 0.001,
+      "step": 18804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.711098768860502e-06,
+      "loss": 0.0005,
+      "step": 18806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.706470424881978e-06,
+      "loss": 1.502,
+      "step": 18808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.701842080903453e-06,
+      "loss": 0.109,
+      "step": 18810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6972137369249285e-06,
+      "loss": 0.458,
+      "step": 18812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.692585392946405e-06,
+      "loss": 2.0564,
+      "step": 18814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.68795704896788e-06,
+      "loss": 4.7744,
+      "step": 18816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.683328704989356e-06,
+      "loss": 7.8842,
+      "step": 18818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.678700361010831e-06,
+      "loss": 2.5351,
+      "step": 18820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6740720170323065e-06,
+      "loss": 1.3294,
+      "step": 18822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.669443673053781e-06,
+      "loss": 0.7107,
+      "step": 18824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6648153290752565e-06,
+      "loss": 1.5544,
+      "step": 18826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.660186985096732e-06,
+      "loss": 0.0011,
+      "step": 18828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.655558641118208e-06,
+      "loss": 0.004,
+      "step": 18830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.650930297139684e-06,
+      "loss": 0.7693,
+      "step": 18832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.646301953161159e-06,
+      "loss": 0.0158,
+      "step": 18834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6416736091826344e-06,
+      "loss": 0.0036,
+      "step": 18836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.63704526520411e-06,
+      "loss": 3.7194,
+      "step": 18838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.632416921225586e-06,
+      "loss": 1.1555,
+      "step": 18840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6277885772470616e-06,
+      "loss": 0.0229,
+      "step": 18842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.623160233268537e-06,
+      "loss": 4.3888,
+      "step": 18844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.618531889290012e-06,
+      "loss": 2.7807,
+      "step": 18846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.613903545311488e-06,
+      "loss": 0.3388,
+      "step": 18848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.609275201332963e-06,
+      "loss": 0.1767,
+      "step": 18850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.6046468573544395e-06,
+      "loss": 5.0098,
+      "step": 18852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.600018513375915e-06,
+      "loss": 4.879,
+      "step": 18854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.59539016939739e-06,
+      "loss": 2.4376,
+      "step": 18856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.590761825418866e-06,
+      "loss": 1.2702,
+      "step": 18858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.586133481440341e-06,
+      "loss": 0.2083,
+      "step": 18860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.581505137461816e-06,
+      "loss": 1.4088,
+      "step": 18862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.576876793483291e-06,
+      "loss": 0.2145,
+      "step": 18864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.572248449504767e-06,
+      "loss": 1.6796,
+      "step": 18866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.567620105526243e-06,
+      "loss": 0.7068,
+      "step": 18868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.562991761547718e-06,
+      "loss": 0.6942,
+      "step": 18870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.558363417569194e-06,
+      "loss": 8.3282,
+      "step": 18872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.553735073590669e-06,
+      "loss": 3.7797,
+      "step": 18874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.549106729612145e-06,
+      "loss": 0.4063,
+      "step": 18876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.544478385633621e-06,
+      "loss": 2.3526,
+      "step": 18878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.539850041655096e-06,
+      "loss": 3.184,
+      "step": 18880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.535221697676572e-06,
+      "loss": 1.9826,
+      "step": 18882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.530593353698047e-06,
+      "loss": 0.1491,
+      "step": 18884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.525965009719523e-06,
+      "loss": 0.3826,
+      "step": 18886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.521336665740999e-06,
+      "loss": 1.8339,
+      "step": 18888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.516708321762474e-06,
+      "loss": 0.0009,
+      "step": 18890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.51207997778395e-06,
+      "loss": 4.9544,
+      "step": 18892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.507451633805425e-06,
+      "loss": 0.1184,
+      "step": 18894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.5028232898269006e-06,
+      "loss": 2.992,
+      "step": 18896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.498194945848375e-06,
+      "loss": 4.9246,
+      "step": 18898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.4935666018698506e-06,
+      "loss": 0.4826,
+      "step": 18900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.488938257891326e-06,
+      "loss": 0.0019,
+      "step": 18902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.484309913912802e-06,
+      "loss": 7.2973,
+      "step": 18904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.479681569934278e-06,
+      "loss": 1.6394,
+      "step": 18906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.475053225955753e-06,
+      "loss": 2.891,
+      "step": 18908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.4704248819772285e-06,
+      "loss": 0.195,
+      "step": 18910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.465796537998704e-06,
+      "loss": 1.1159,
+      "step": 18912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.46116819402018e-06,
+      "loss": 1.8886,
+      "step": 18914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.456539850041656e-06,
+      "loss": 0.0477,
+      "step": 18916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.451911506063131e-06,
+      "loss": 0.6792,
+      "step": 18918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.4472831620846065e-06,
+      "loss": 2.2394,
+      "step": 18920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.442654818106082e-06,
+      "loss": 0.0011,
+      "step": 18922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.438026474127558e-06,
+      "loss": 3.6483,
+      "step": 18924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.433398130149034e-06,
+      "loss": 0.0014,
+      "step": 18926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.428769786170509e-06,
+      "loss": 2.8933,
+      "step": 18928
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.4241414421919845e-06,
+      "loss": 3.9278,
+      "step": 18930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.41951309821346e-06,
+      "loss": 0.0282,
+      "step": 18932
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.4148847542349345e-06,
+      "loss": 0.0015,
+      "step": 18934
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.41025641025641e-06,
+      "loss": 0.0077,
+      "step": 18936
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.405628066277885e-06,
+      "loss": 0.0031,
+      "step": 18938
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.400999722299361e-06,
+      "loss": 2.6474,
+      "step": 18940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.396371378320837e-06,
+      "loss": 0.0033,
+      "step": 18942
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.3917430343423125e-06,
+      "loss": 1.3528,
+      "step": 18944
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.387114690363788e-06,
+      "loss": 1.7129,
+      "step": 18946
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.382486346385263e-06,
+      "loss": 1.3138,
+      "step": 18948
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.377858002406739e-06,
+      "loss": 1.4547,
+      "step": 18950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.373229658428215e-06,
+      "loss": 1.5437,
+      "step": 18952
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.3686013144496904e-06,
+      "loss": 0.6125,
+      "step": 18954
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.363972970471166e-06,
+      "loss": 0.7609,
+      "step": 18956
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.359344626492641e-06,
+      "loss": 1.2146,
+      "step": 18958
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.354716282514117e-06,
+      "loss": 2.8508,
+      "step": 18960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.350087938535593e-06,
+      "loss": 1.7008,
+      "step": 18962
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.345459594557068e-06,
+      "loss": 0.3529,
+      "step": 18964
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.340831250578544e-06,
+      "loss": 1.0015,
+      "step": 18966
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.336202906600019e-06,
+      "loss": 0.0032,
+      "step": 18968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.331574562621494e-06,
+      "loss": 1.5365,
+      "step": 18970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.326946218642969e-06,
+      "loss": 0.6966,
+      "step": 18972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.322317874664445e-06,
+      "loss": 0.5463,
+      "step": 18974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.31768953068592e-06,
+      "loss": 2.3454,
+      "step": 18976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.313061186707396e-06,
+      "loss": 1.1663,
+      "step": 18978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.308432842728872e-06,
+      "loss": 0.1678,
+      "step": 18980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.303804498750347e-06,
+      "loss": 0.0009,
+      "step": 18982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.299176154771823e-06,
+      "loss": 1.4845,
+      "step": 18984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.294547810793298e-06,
+      "loss": 0.7664,
+      "step": 18986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.289919466814774e-06,
+      "loss": 1.2494,
+      "step": 18988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.28529112283625e-06,
+      "loss": 0.5851,
+      "step": 18990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.280662778857725e-06,
+      "loss": 0.0007,
+      "step": 18992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.276034434879201e-06,
+      "loss": 1.1431,
+      "step": 18994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.271406090900676e-06,
+      "loss": 1.4807,
+      "step": 18996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.266777746922152e-06,
+      "loss": 2.4738,
+      "step": 18998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.262149402943628e-06,
+      "loss": 2.6152,
+      "step": 19000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.257521058965103e-06,
+      "loss": 0.127,
+      "step": 19002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.252892714986579e-06,
+      "loss": 2.0493,
+      "step": 19004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.248264371008054e-06,
+      "loss": 0.0022,
+      "step": 19006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.2436360270295294e-06,
+      "loss": 2.2907,
+      "step": 19008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.239007683051005e-06,
+      "loss": 0.7032,
+      "step": 19010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.23437933907248e-06,
+      "loss": 0.0166,
+      "step": 19012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.229750995093956e-06,
+      "loss": 0.7407,
+      "step": 19014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.225122651115431e-06,
+      "loss": 1.8433,
+      "step": 19016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.2204943071369066e-06,
+      "loss": 4.0245,
+      "step": 19018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.215865963158382e-06,
+      "loss": 1.7886,
+      "step": 19020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.211237619179857e-06,
+      "loss": 3.7121,
+      "step": 19022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.206609275201334e-06,
+      "loss": 3.7429,
+      "step": 19024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.201980931222809e-06,
+      "loss": 4.7632,
+      "step": 19026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.1973525872442845e-06,
+      "loss": 0.0037,
+      "step": 19028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.19272424326576e-06,
+      "loss": 0.0035,
+      "step": 19030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.188095899287235e-06,
+      "loss": 2.5502,
+      "step": 19032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.183467555308711e-06,
+      "loss": 0.3566,
+      "step": 19034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.178839211330186e-06,
+      "loss": 0.6519,
+      "step": 19036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.174210867351662e-06,
+      "loss": 0.7912,
+      "step": 19038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.169582523373137e-06,
+      "loss": 7.0106,
+      "step": 19040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.164954179394613e-06,
+      "loss": 1.8813,
+      "step": 19042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.160325835416089e-06,
+      "loss": 3.9019,
+      "step": 19044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.155697491437564e-06,
+      "loss": 4.254,
+      "step": 19046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.15106914745904e-06,
+      "loss": 0.0005,
+      "step": 19048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.146440803480515e-06,
+      "loss": 0.6391,
+      "step": 19050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.1418124595019905e-06,
+      "loss": 1.0627,
+      "step": 19052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.137184115523466e-06,
+      "loss": 0.6827,
+      "step": 19054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.132555771544941e-06,
+      "loss": 1.2934,
+      "step": 19056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.127927427566417e-06,
+      "loss": 0.0024,
+      "step": 19058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.123299083587892e-06,
+      "loss": 1.5505,
+      "step": 19060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.1186707396093684e-06,
+      "loss": 0.5083,
+      "step": 19062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.114042395630844e-06,
+      "loss": 0.2192,
+      "step": 19064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.109414051652319e-06,
+      "loss": 0.0181,
+      "step": 19066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.104785707673795e-06,
+      "loss": 1.0363,
+      "step": 19068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.10015736369527e-06,
+      "loss": 1.0302,
+      "step": 19070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.0955290197167456e-06,
+      "loss": 0.0192,
+      "step": 19072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.090900675738221e-06,
+      "loss": 1.0003,
+      "step": 19074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.086272331759696e-06,
+      "loss": 0.5009,
+      "step": 19076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.081643987781172e-06,
+      "loss": 1.8941,
+      "step": 19078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.077015643802648e-06,
+      "loss": 0.0005,
+      "step": 19080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.0723872998241235e-06,
+      "loss": 0.2932,
+      "step": 19082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.067758955845599e-06,
+      "loss": 0.1687,
+      "step": 19084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.063130611867074e-06,
+      "loss": 3.025,
+      "step": 19086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.05850226788855e-06,
+      "loss": 1.6565,
+      "step": 19088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.053873923910025e-06,
+      "loss": 0.001,
+      "step": 19090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.049245579931501e-06,
+      "loss": 4.5432,
+      "step": 19092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.044617235952976e-06,
+      "loss": 0.5734,
+      "step": 19094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.0399888919744515e-06,
+      "loss": 3.3936,
+      "step": 19096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.035360547995928e-06,
+      "loss": 1.1206,
+      "step": 19098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.030732204017403e-06,
+      "loss": 2.2224,
+      "step": 19100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.026103860038879e-06,
+      "loss": 1.0426,
+      "step": 19102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.021475516060354e-06,
+      "loss": 4.089,
+      "step": 19104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.0168471720818295e-06,
+      "loss": 0.0153,
+      "step": 19106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.012218828103305e-06,
+      "loss": 3.1542,
+      "step": 19108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.00759048412478e-06,
+      "loss": 0.0003,
+      "step": 19110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6.002962140146256e-06,
+      "loss": 0.0008,
+      "step": 19112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.998333796167731e-06,
+      "loss": 1.9317,
+      "step": 19114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9937054521892074e-06,
+      "loss": 3.1706,
+      "step": 19116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.989077108210683e-06,
+      "loss": 2.2755,
+      "step": 19118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.984448764232158e-06,
+      "loss": 1.7302,
+      "step": 19120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.979820420253634e-06,
+      "loss": 2.4883,
+      "step": 19122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.975192076275109e-06,
+      "loss": 1.9437,
+      "step": 19124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9705637322965846e-06,
+      "loss": 1.5467,
+      "step": 19126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.96593538831806e-06,
+      "loss": 0.9198,
+      "step": 19128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.961307044339535e-06,
+      "loss": 4.4292,
+      "step": 19130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956678700361011e-06,
+      "loss": 3.7091,
+      "step": 19132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.952050356382487e-06,
+      "loss": 0.0082,
+      "step": 19134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9474220124039625e-06,
+      "loss": 2.4708,
+      "step": 19136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.942793668425438e-06,
+      "loss": 1.3665,
+      "step": 19138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.938165324446913e-06,
+      "loss": 4.0494,
+      "step": 19140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.933536980468388e-06,
+      "loss": 0.1515,
+      "step": 19142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.928908636489864e-06,
+      "loss": 0.0077,
+      "step": 19144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.92428029251134e-06,
+      "loss": 2.1912,
+      "step": 19146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.919651948532815e-06,
+      "loss": 0.4103,
+      "step": 19148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9150236045542905e-06,
+      "loss": 0.4467,
+      "step": 19150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.910395260575766e-06,
+      "loss": 1.2256,
+      "step": 19152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.905766916597242e-06,
+      "loss": 0.0327,
+      "step": 19154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.901138572618718e-06,
+      "loss": 3.3413,
+      "step": 19156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.896510228640193e-06,
+      "loss": 0.0011,
+      "step": 19158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.891881884661668e-06,
+      "loss": 1.0851,
+      "step": 19160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.887253540683144e-06,
+      "loss": 0.9687,
+      "step": 19162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.882625196704619e-06,
+      "loss": 1.4474,
+      "step": 19164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.877996852726095e-06,
+      "loss": 0.0063,
+      "step": 19166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.87336850874757e-06,
+      "loss": 1.552,
+      "step": 19168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.868740164769046e-06,
+      "loss": 0.2935,
+      "step": 19170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.864111820790522e-06,
+      "loss": 3.1405,
+      "step": 19172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.859483476811997e-06,
+      "loss": 1.8126,
+      "step": 19174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.854855132833473e-06,
+      "loss": 0.0015,
+      "step": 19176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.850226788854947e-06,
+      "loss": 3.2565,
+      "step": 19178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8455984448764236e-06,
+      "loss": 0.342,
+      "step": 19180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.840970100897899e-06,
+      "loss": 0.0185,
+      "step": 19182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.836341756919374e-06,
+      "loss": 3.5792,
+      "step": 19184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.83171341294085e-06,
+      "loss": 5.2717,
+      "step": 19186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.827085068962325e-06,
+      "loss": 0.6778,
+      "step": 19188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8224567249838015e-06,
+      "loss": 0.7874,
+      "step": 19190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.817828381005277e-06,
+      "loss": 0.0111,
+      "step": 19192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.813200037026752e-06,
+      "loss": 0.483,
+      "step": 19194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.808571693048227e-06,
+      "loss": 3.5929,
+      "step": 19196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.803943349069703e-06,
+      "loss": 0.0006,
+      "step": 19198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.799315005091179e-06,
+      "loss": 0.2276,
+      "step": 19200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.794686661112654e-06,
+      "loss": 0.0815,
+      "step": 19202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7900583171341295e-06,
+      "loss": 0.4711,
+      "step": 19204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.785429973155605e-06,
+      "loss": 0.9405,
+      "step": 19206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.780801629177081e-06,
+      "loss": 4.6065,
+      "step": 19208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.776173285198557e-06,
+      "loss": 9.458,
+      "step": 19210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.771544941220032e-06,
+      "loss": 9.4888,
+      "step": 19212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.766916597241507e-06,
+      "loss": 1.5839,
+      "step": 19214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.762288253262983e-06,
+      "loss": 4.6053,
+      "step": 19216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.757659909284458e-06,
+      "loss": 0.2925,
+      "step": 19218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.753031565305934e-06,
+      "loss": 0.2319,
+      "step": 19220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.748403221327409e-06,
+      "loss": 0.0886,
+      "step": 19222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.743774877348885e-06,
+      "loss": 1.1154,
+      "step": 19224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.739146533370361e-06,
+      "loss": 1.3807,
+      "step": 19226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.734518189391836e-06,
+      "loss": 0.6781,
+      "step": 19228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.729889845413312e-06,
+      "loss": 1.3797,
+      "step": 19230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.725261501434786e-06,
+      "loss": 0.2657,
+      "step": 19232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7206331574562626e-06,
+      "loss": 0.0068,
+      "step": 19234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.716004813477738e-06,
+      "loss": 0.4468,
+      "step": 19236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.711376469499213e-06,
+      "loss": 1.6642,
+      "step": 19238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.706748125520689e-06,
+      "loss": 0.0024,
+      "step": 19240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.702119781542164e-06,
+      "loss": 1.6819,
+      "step": 19242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.6974914375636405e-06,
+      "loss": 4.1821,
+      "step": 19244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.692863093585116e-06,
+      "loss": 2.0712,
+      "step": 19246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.688234749606591e-06,
+      "loss": 2.4685,
+      "step": 19248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.683606405628066e-06,
+      "loss": 0.0416,
+      "step": 19250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.678978061649541e-06,
+      "loss": 0.2607,
+      "step": 19252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.674349717671018e-06,
+      "loss": 1.4553,
+      "step": 19254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.669721373692493e-06,
+      "loss": 3.8406,
+      "step": 19256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.6650930297139685e-06,
+      "loss": 2.2556,
+      "step": 19258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.660464685735444e-06,
+      "loss": 4.7357,
+      "step": 19260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.655836341756919e-06,
+      "loss": 2.3174,
+      "step": 19262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.651207997778396e-06,
+      "loss": 0.0117,
+      "step": 19264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.646579653799871e-06,
+      "loss": 1.4404,
+      "step": 19266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.6419513098213465e-06,
+      "loss": 1.2218,
+      "step": 19268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.637322965842821e-06,
+      "loss": 0.0028,
+      "step": 19270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.632694621864297e-06,
+      "loss": 1.1307,
+      "step": 19272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.628066277885773e-06,
+      "loss": 1.0859,
+      "step": 19274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.623437933907248e-06,
+      "loss": 1.8234,
+      "step": 19276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.618809589928724e-06,
+      "loss": 0.0016,
+      "step": 19278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.614181245950199e-06,
+      "loss": 0.0516,
+      "step": 19280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.609552901971675e-06,
+      "loss": 1.628,
+      "step": 19282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.604924557993151e-06,
+      "loss": 4.9162,
+      "step": 19284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.600296214014626e-06,
+      "loss": 1.837,
+      "step": 19286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.595667870036101e-06,
+      "loss": 1.5987,
+      "step": 19288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.591039526057577e-06,
+      "loss": 1.7679,
+      "step": 19290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.5864111820790524e-06,
+      "loss": 0.0007,
+      "step": 19292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.581782838100528e-06,
+      "loss": 0.0141,
+      "step": 19294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.577154494122003e-06,
+      "loss": 0.8932,
+      "step": 19296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.572526150143479e-06,
+      "loss": 1.3604,
+      "step": 19298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.567897806164955e-06,
+      "loss": 0.8716,
+      "step": 19300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.56326946218643e-06,
+      "loss": 0.0008,
+      "step": 19302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.558641118207906e-06,
+      "loss": 1.8313,
+      "step": 19304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.55401277422938e-06,
+      "loss": 0.0031,
+      "step": 19306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.549384430250857e-06,
+      "loss": 1.8895,
+      "step": 19308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.544756086272332e-06,
+      "loss": 0.0029,
+      "step": 19310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.5401277422938075e-06,
+      "loss": 0.0004,
+      "step": 19312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.535499398315283e-06,
+      "loss": 0.1039,
+      "step": 19314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.530871054336758e-06,
+      "loss": 1.9743,
+      "step": 19316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.526242710358235e-06,
+      "loss": 1.329,
+      "step": 19318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.52161436637971e-06,
+      "loss": 0.0015,
+      "step": 19320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.5169860224011855e-06,
+      "loss": 1.1758,
+      "step": 19322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.51235767842266e-06,
+      "loss": 0.0099,
+      "step": 19324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.507729334444136e-06,
+      "loss": 1.1392,
+      "step": 19326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.503100990465612e-06,
+      "loss": 0.5247,
+      "step": 19328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.498472646487087e-06,
+      "loss": 1.6777,
+      "step": 19330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.493844302508563e-06,
+      "loss": 0.2181,
+      "step": 19332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.489215958530038e-06,
+      "loss": 0.4036,
+      "step": 19334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.484587614551514e-06,
+      "loss": 5.3025,
+      "step": 19336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.47995927057299e-06,
+      "loss": 3.5041,
+      "step": 19338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.475330926594465e-06,
+      "loss": 1.4783,
+      "step": 19340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.47070258261594e-06,
+      "loss": 5.237,
+      "step": 19342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.466074238637415e-06,
+      "loss": 1.0331,
+      "step": 19344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4614458946588914e-06,
+      "loss": 1.8942,
+      "step": 19346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.456817550680367e-06,
+      "loss": 0.6359,
+      "step": 19348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.452189206701842e-06,
+      "loss": 0.8661,
+      "step": 19350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.447560862723318e-06,
+      "loss": 1.5262,
+      "step": 19352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.442932518744793e-06,
+      "loss": 0.0064,
+      "step": 19354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.438304174766269e-06,
+      "loss": 0.0146,
+      "step": 19356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.433675830787745e-06,
+      "loss": 2.0817,
+      "step": 19358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.429047486809219e-06,
+      "loss": 5.4409,
+      "step": 19360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.424419142830695e-06,
+      "loss": 3.9706,
+      "step": 19362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.419790798852171e-06,
+      "loss": 1.0827,
+      "step": 19364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.4151624548736465e-06,
+      "loss": 2.0929,
+      "step": 19366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.410534110895122e-06,
+      "loss": 1.2173,
+      "step": 19368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.405905766916597e-06,
+      "loss": 4.6172,
+      "step": 19370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.401277422938073e-06,
+      "loss": 0.0012,
+      "step": 19372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.396649078959549e-06,
+      "loss": 1.14,
+      "step": 19374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.3920207349810245e-06,
+      "loss": 5.6752,
+      "step": 19376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.387392391002499e-06,
+      "loss": 0.9607,
+      "step": 19378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.3827640470239745e-06,
+      "loss": 0.6249,
+      "step": 19380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.378135703045451e-06,
+      "loss": 0.0709,
+      "step": 19382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.373507359066926e-06,
+      "loss": 4.3825,
+      "step": 19384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.368879015088402e-06,
+      "loss": 0.1657,
+      "step": 19386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.364250671109877e-06,
+      "loss": 0.8832,
+      "step": 19388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.3596223271313525e-06,
+      "loss": 2.1735,
+      "step": 19390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.354993983152829e-06,
+      "loss": 1.4918,
+      "step": 19392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.350365639174304e-06,
+      "loss": 2.7286,
+      "step": 19394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.345737295195779e-06,
+      "loss": 0.716,
+      "step": 19396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.341108951217254e-06,
+      "loss": 0.0034,
+      "step": 19398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.3364806072387304e-06,
+      "loss": 0.0018,
+      "step": 19400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.331852263260206e-06,
+      "loss": 2.6572,
+      "step": 19402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.327223919281681e-06,
+      "loss": 0.302,
+      "step": 19404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.322595575303157e-06,
+      "loss": 0.5596,
+      "step": 19406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.317967231324632e-06,
+      "loss": 2.118,
+      "step": 19408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.313338887346108e-06,
+      "loss": 1.7039,
+      "step": 19410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.308710543367584e-06,
+      "loss": 0.259,
+      "step": 19412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.304082199389058e-06,
+      "loss": 0.359,
+      "step": 19414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.299453855410534e-06,
+      "loss": 2.8029,
+      "step": 19416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.29482551143201e-06,
+      "loss": 1.1135,
+      "step": 19418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.2901971674534855e-06,
+      "loss": 0.7646,
+      "step": 19420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.285568823474961e-06,
+      "loss": 0.2531,
+      "step": 19422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.280940479496436e-06,
+      "loss": 2.029,
+      "step": 19424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.276312135517912e-06,
+      "loss": 0.5372,
+      "step": 19426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.271683791539388e-06,
+      "loss": 2.4706,
+      "step": 19428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.2670554475608635e-06,
+      "loss": 6.8244,
+      "step": 19430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.262427103582338e-06,
+      "loss": 1.6946,
+      "step": 19432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.2577987596038135e-06,
+      "loss": 0.5327,
+      "step": 19434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.25317041562529e-06,
+      "loss": 6.9696,
+      "step": 19436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.248542071646765e-06,
+      "loss": 6.123,
+      "step": 19438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.243913727668241e-06,
+      "loss": 0.0011,
+      "step": 19440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.239285383689716e-06,
+      "loss": 0.0047,
+      "step": 19442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.2346570397111915e-06,
+      "loss": 1.0766,
+      "step": 19444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.230028695732668e-06,
+      "loss": 1.3952,
+      "step": 19446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.225400351754143e-06,
+      "loss": 0.5235,
+      "step": 19448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.220772007775618e-06,
+      "loss": 0.1963,
+      "step": 19450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.216143663797093e-06,
+      "loss": 3.9422,
+      "step": 19452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.211515319818569e-06,
+      "loss": 1.3211,
+      "step": 19454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.206886975840045e-06,
+      "loss": 3.315,
+      "step": 19456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.20225863186152e-06,
+      "loss": 3.8378,
+      "step": 19458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.197630287882996e-06,
+      "loss": 1.0021,
+      "step": 19460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.193001943904471e-06,
+      "loss": 1.9531,
+      "step": 19462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1883735999259466e-06,
+      "loss": 3.4466,
+      "step": 19464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.183745255947423e-06,
+      "loss": 4.8198,
+      "step": 19466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.179116911968897e-06,
+      "loss": 0.7028,
+      "step": 19468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.174488567990373e-06,
+      "loss": 3.145,
+      "step": 19470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.169860224011848e-06,
+      "loss": 0.8533,
+      "step": 19472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1652318800333245e-06,
+      "loss": 0.4271,
+      "step": 19474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1606035360548e-06,
+      "loss": 1.2251,
+      "step": 19476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.155975192076275e-06,
+      "loss": 6.8359,
+      "step": 19478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.151346848097751e-06,
+      "loss": 0.941,
+      "step": 19480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.146718504119226e-06,
+      "loss": 1.8861,
+      "step": 19482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1420901601407025e-06,
+      "loss": 0.9458,
+      "step": 19484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.137461816162177e-06,
+      "loss": 0.0973,
+      "step": 19486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1328334721836525e-06,
+      "loss": 1.2241,
+      "step": 19488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.9336,
+      "step": 19490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.123576784226604e-06,
+      "loss": 1.8695,
+      "step": 19492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.11894844024808e-06,
+      "loss": 0.0037,
+      "step": 19494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.114320096269555e-06,
+      "loss": 5.0138,
+      "step": 19496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.1096917522910305e-06,
+      "loss": 0.0034,
+      "step": 19498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.105063408312506e-06,
+      "loss": 1.4566,
+      "step": 19500
+    }
+  ],
+  "max_steps": 21706,
+  "num_train_epochs": 1,
+  "total_flos": 762615419904000.0,
+  "trial_name": null,
+  "trial_params": null
+}