diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7594 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 2523,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.6576,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.6592,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 0.6936,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 0.6209,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 0.7266,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 0.583,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0,
+      "loss": 0.5872,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0,
+      "loss": 0.6971,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0,
+      "loss": 0.6075,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.6744,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.5986,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6547,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.5527,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.4789,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.5411,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.5627,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.5674,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2e-05,
+      "loss": 0.5058,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5078,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 0.5258,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.5611,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.4715,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.7397,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.5526,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999996823967381e-05,
+      "loss": 0.5387,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999872958896982e-05,
+      "loss": 0.5412,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999714158274743e-05,
+      "loss": 0.487,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999491838815805e-05,
+      "loss": 0.5345,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999920600193236e-05,
+      "loss": 0.5062,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998856649440058e-05,
+      "loss": 0.5077,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999844378355801e-05,
+      "loss": 0.5029,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999796740690877e-05,
+      "loss": 0.5374,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997427522518315e-05,
+      "loss": 0.5321,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999682413381602e-05,
+      "loss": 0.4695,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996157244634647e-05,
+      "loss": 0.4798,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9995799988672346e-05,
+      "loss": 0.4682,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999503785684114e-05,
+      "loss": 0.5138,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9994212235877407e-05,
+      "loss": 0.5271,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999332313102555e-05,
+      "loss": 0.5526,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999237054793322e-05,
+      "loss": 0.5385,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9991354492651283e-05,
+      "loss": 0.5259,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9990274971633787e-05,
+      "loss": 0.5054,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9989131991737928e-05,
+      "loss": 0.4698,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998792556022398e-05,
+      "loss": 0.4909,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.998665568475528e-05,
+      "loss": 0.462,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.998532237339816e-05,
+      "loss": 0.4898,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9983925634621894e-05,
+      "loss": 0.4771,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.998246547729867e-05,
+      "loss": 0.4774,
+      "step": 96
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.998094191070349e-05,
+      "loss": 0.421,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.997935494451416e-05,
+      "loss": 0.4991,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9977704588811183e-05,
+      "loss": 0.5349,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9975990854077733e-05,
+      "loss": 0.4717,
+      "step": 104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9974213751199556e-05,
+      "loss": 0.5024,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9972373291464933e-05,
+      "loss": 0.5904,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9970469486564585e-05,
+      "loss": 0.4817,
+      "step": 110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99685023485916e-05,
+      "loss": 0.488,
+      "step": 112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9966471890041374e-05,
+      "loss": 0.5262,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9964378123811502e-05,
+      "loss": 0.4721,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9962221063201734e-05,
+      "loss": 0.4594,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.996000072191385e-05,
+      "loss": 0.5068,
+      "step": 120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9957717114051608e-05,
+      "loss": 0.5169,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9955370254120635e-05,
+      "loss": 0.4781,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9952960157028335e-05,
+      "loss": 0.4682,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9950486838083807e-05,
+      "loss": 0.4754,
+      "step": 128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.994795031299773e-05,
+      "loss": 0.479,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9945350597882275e-05,
+      "loss": 0.5381,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9942687709251006e-05,
+      "loss": 0.5457,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.993996166401877e-05,
+      "loss": 0.4653,
+      "step": 136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9937172479501573e-05,
+      "loss": 0.4686,
+      "step": 138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9934320173416502e-05,
+      "loss": 0.5098,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9931404763881598e-05,
+      "loss": 0.4823,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.992842626941572e-05,
+      "loss": 0.4999,
+      "step": 144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.992538470893847e-05,
+      "loss": 0.4868,
+      "step": 146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.992228010177003e-05,
+      "loss": 0.4668,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9919112467631074e-05,
+      "loss": 0.4955,
+      "step": 150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.991588182664262e-05,
+      "loss": 0.4855,
+      "step": 152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99125881993259e-05,
+      "loss": 0.4646,
+      "step": 154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9909231606602266e-05,
+      "loss": 0.5182,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9905812069793002e-05,
+      "loss": 0.4857,
+      "step": 158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.990232961061924e-05,
+      "loss": 0.5048,
+      "step": 160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.989878425120177e-05,
+      "loss": 0.4597,
+      "step": 162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9895176014060964e-05,
+      "loss": 0.4387,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9891504922116572e-05,
+      "loss": 0.4556,
+      "step": 166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9887770998687607e-05,
+      "loss": 0.4937,
+      "step": 168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9883974267492202e-05,
+      "loss": 0.5738,
+      "step": 170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9880114752647434e-05,
+      "loss": 0.4769,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9876192478669197e-05,
+      "loss": 0.5109,
+      "step": 174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.987220747047203e-05,
+      "loss": 0.4926,
+      "step": 176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9868159753368964e-05,
+      "loss": 0.4726,
+      "step": 178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9864049353071365e-05,
+      "loss": 0.4675,
+      "step": 180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.985987629568876e-05,
+      "loss": 0.4723,
+      "step": 182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9855640607728684e-05,
+      "loss": 0.4215,
+      "step": 184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9851342316096503e-05,
+      "loss": 0.505,
+      "step": 186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.984698144809525e-05,
+      "loss": 0.5278,
+      "step": 188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9842558031425434e-05,
+      "loss": 0.4836,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.983807209418489e-05,
+      "loss": 0.445,
+      "step": 192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9833523664868587e-05,
+      "loss": 0.4585,
+      "step": 194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.982891277236845e-05,
+      "loss": 0.4934,
+      "step": 196
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982423944597315e-05,
+      "loss": 0.4899,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981950371536798e-05,
+      "loss": 0.5321,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9814705610634602e-05,
+      "loss": 0.4883,
+      "step": 202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980984516225089e-05,
+      "loss": 0.4805,
+      "step": 204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9804922401090732e-05,
+      "loss": 0.4507,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9799937358423826e-05,
+      "loss": 0.4872,
+      "step": 208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9794890065915486e-05,
+      "loss": 0.5495,
+      "step": 210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9789780555626444e-05,
+      "loss": 0.4936,
+      "step": 212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9784608860012652e-05,
+      "loss": 0.4769,
+      "step": 214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9779375011925046e-05,
+      "loss": 0.4714,
+      "step": 216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9774079044609373e-05,
+      "loss": 0.5064,
+      "step": 218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.976872099170597e-05,
+      "loss": 0.4629,
+      "step": 220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.976330088724953e-05,
+      "loss": 0.4149,
+      "step": 222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9757818765668916e-05,
+      "loss": 0.4565,
+      "step": 224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9752274661786916e-05,
+      "loss": 0.4233,
+      "step": 226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9746668610820047e-05,
+      "loss": 0.4557,
+      "step": 228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9741000648378303e-05,
+      "loss": 0.5186,
+      "step": 230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9735270810464958e-05,
+      "loss": 0.4425,
+      "step": 232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.972947913347631e-05,
+      "loss": 0.5274,
+      "step": 234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9723625654201472e-05,
+      "loss": 0.4087,
+      "step": 236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.971771040982213e-05,
+      "loss": 0.4412,
+      "step": 238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9711733437912293e-05,
+      "loss": 0.4998,
+      "step": 240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9705694776438084e-05,
+      "loss": 0.5343,
+      "step": 242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9699594463757475e-05,
+      "loss": 0.4365,
+      "step": 244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9693432538620046e-05,
+      "loss": 0.4877,
+      "step": 246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9687209040166748e-05,
+      "loss": 0.4361,
+      "step": 248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.968092400792965e-05,
+      "loss": 0.4555,
+      "step": 250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.967457748183169e-05,
+      "loss": 0.4353,
+      "step": 252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.966816950218641e-05,
+      "loss": 0.3476,
+      "step": 254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9661700109697718e-05,
+      "loss": 0.4342,
+      "step": 256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9655169345459622e-05,
+      "loss": 0.4603,
+      "step": 258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.964857725095595e-05,
+      "loss": 0.4514,
+      "step": 260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.964192386806013e-05,
+      "loss": 0.4889,
+      "step": 262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9635209239034872e-05,
+      "loss": 0.5255,
+      "step": 264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.962843340653195e-05,
+      "loss": 0.5325,
+      "step": 266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9621596413591885e-05,
+      "loss": 0.5163,
+      "step": 268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96146983036437e-05,
+      "loss": 0.3708,
+      "step": 270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.960773912050465e-05,
+      "loss": 0.5378,
+      "step": 272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.960071890837991e-05,
+      "loss": 0.4389,
+      "step": 274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9593637711862335e-05,
+      "loss": 0.4565,
+      "step": 276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9586495575932137e-05,
+      "loss": 0.5446,
+      "step": 278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.957929254595664e-05,
+      "loss": 0.5079,
+      "step": 280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.957202866768995e-05,
+      "loss": 0.3997,
+      "step": 282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9564703987272703e-05,
+      "loss": 0.4934,
+      "step": 284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9557318551231745e-05,
+      "loss": 0.5076,
+      "step": 286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9549872406479843e-05,
+      "loss": 0.4487,
+      "step": 288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9542365600315403e-05,
+      "loss": 0.5515,
+      "step": 290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.953479818042214e-05,
+      "loss": 0.5262,
+      "step": 292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.95271701948688e-05,
+      "loss": 0.4953,
+      "step": 294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.951948169210885e-05,
+      "loss": 0.5127,
+      "step": 296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9511732720980156e-05,
+      "loss": 0.4796,
+      "step": 298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.950392333070469e-05,
+      "loss": 0.5016,
+      "step": 300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9496053570888205e-05,
+      "loss": 0.5114,
+      "step": 302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9488123491519935e-05,
+      "loss": 0.4471,
+      "step": 304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9480133142972257e-05,
+      "loss": 0.4427,
+      "step": 306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.947208257600039e-05,
+      "loss": 0.457,
+      "step": 308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9463971841742057e-05,
+      "loss": 0.5193,
+      "step": 310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.945580099171717e-05,
+      "loss": 0.5337,
+      "step": 312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9447570077827503e-05,
+      "loss": 0.4758,
+      "step": 314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9439279152356363e-05,
+      "loss": 0.4223,
+      "step": 316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.943092826796824e-05,
+      "loss": 0.4909,
+      "step": 318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9422517477708506e-05,
+      "loss": 0.4615,
+      "step": 320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9414046835003043e-05,
+      "loss": 0.3863,
+      "step": 322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.940551639365793e-05,
+      "loss": 0.4676,
+      "step": 324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.488,
+      "step": 326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.938827633217193e-05,
+      "loss": 0.4888,
+      "step": 328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9379566821541034e-05,
+      "loss": 0.4754,
+      "step": 330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9370797731289784e-05,
+      "loss": 0.7974,
+      "step": 332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.936196911712001e-05,
+      "loss": 0.3944,
+      "step": 334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9353081035111644e-05,
+      "loss": 0.4883,
+      "step": 336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9344133541722368e-05,
+      "loss": 0.5429,
+      "step": 338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9335126693787237e-05,
+      "loss": 0.4573,
+      "step": 340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9326060548518342e-05,
+      "loss": 0.4276,
+      "step": 342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9316935163504424e-05,
+      "loss": 0.5089,
+      "step": 344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.930775059671053e-05,
+      "loss": 0.4285,
+      "step": 346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9298506906477623e-05,
+      "loss": 0.4438,
+      "step": 348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9289204151522227e-05,
+      "loss": 0.4644,
+      "step": 350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.927984239093605e-05,
+      "loss": 0.3904,
+      "step": 352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9270421684185603e-05,
+      "loss": 0.5486,
+      "step": 354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9260942091111836e-05,
+      "loss": 0.5429,
+      "step": 356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9251403671929738e-05,
+      "loss": 0.4597,
+      "step": 358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9241806487227967e-05,
+      "loss": 0.4721,
+      "step": 360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923215059796847e-05,
+      "loss": 0.4689,
+      "step": 362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922243606548609e-05,
+      "loss": 0.4716,
+      "step": 364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9212662951488162e-05,
+      "loss": 0.4993,
+      "step": 366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9202831318054153e-05,
+      "loss": 0.4557,
+      "step": 368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9192941227635232e-05,
+      "loss": 0.4701,
+      "step": 370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.91829927430539e-05,
+      "loss": 0.4032,
+      "step": 372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9172985927503584e-05,
+      "loss": 0.4743,
+      "step": 374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9162920844548227e-05,
+      "loss": 0.4491,
+      "step": 376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9152797558121894e-05,
+      "loss": 0.429,
+      "step": 378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9142616132528356e-05,
+      "loss": 0.4707,
+      "step": 380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.91323766324407e-05,
+      "loss": 0.4371,
+      "step": 382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.912207912290089e-05,
+      "loss": 0.5172,
+      "step": 384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9111723669319385e-05,
+      "loss": 0.4482,
+      "step": 386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.91013103374747e-05,
+      "loss": 0.4701,
+      "step": 388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9090839193513e-05,
+      "loss": 0.4737,
+      "step": 390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9080310303947668e-05,
+      "loss": 0.4922,
+      "step": 392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9069723735658903e-05,
+      "loss": 0.4081,
+      "step": 394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9059079555893277e-05,
+      "loss": 0.486,
+      "step": 396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9048377832263314e-05,
+      "loss": 0.4674,
+      "step": 398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.903761863274706e-05,
+      "loss": 0.4528,
+      "step": 400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.902680202568765e-05,
+      "loss": 0.4716,
+      "step": 402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9015928079792884e-05,
+      "loss": 0.5213,
+      "step": 404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9004996864134767e-05,
+      "loss": 0.4836,
+      "step": 406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8994008448149103e-05,
+      "loss": 0.5513,
+      "step": 408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8982962901635022e-05,
+      "loss": 0.4634,
+      "step": 410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8971860294754554e-05,
+      "loss": 0.5111,
+      "step": 412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8960700698032194e-05,
+      "loss": 0.4474,
+      "step": 414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.894948418235441e-05,
+      "loss": 0.4785,
+      "step": 416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8938210818969257e-05,
+      "loss": 0.4977,
+      "step": 418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8926880679485865e-05,
+      "loss": 0.4346,
+      "step": 420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8915493835874026e-05,
+      "loss": 0.4139,
+      "step": 422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8904050360463708e-05,
+      "loss": 0.4975,
+      "step": 424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8892550325944617e-05,
+      "loss": 0.4766,
+      "step": 426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.888099380536572e-05,
+      "loss": 0.484,
+      "step": 428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.886938087213479e-05,
+      "loss": 0.4301,
+      "step": 430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.885771160001794e-05,
+      "loss": 0.4474,
+      "step": 432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8845986063139144e-05,
+      "loss": 0.4445,
+      "step": 434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8834204335979777e-05,
+      "loss": 0.4422,
+      "step": 436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8822366493378143e-05,
+      "loss": 0.5337,
+      "step": 438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8810472610528987e-05,
+      "loss": 0.4704,
+      "step": 440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8798522762983026e-05,
+      "loss": 0.458,
+      "step": 442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8786517026646474e-05,
+      "loss": 0.3866,
+      "step": 444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8774455477780557e-05,
+      "loss": 0.4939,
+      "step": 446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8762338193001013e-05,
+      "loss": 0.496,
+      "step": 448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8750165249277625e-05,
+      "loss": 0.4171,
+      "step": 450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.873793672393373e-05,
+      "loss": 0.4582,
+      "step": 452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8725652694645714e-05,
+      "loss": 0.4762,
+      "step": 454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.871331323944254e-05,
+      "loss": 0.3859,
+      "step": 456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8700918436705226e-05,
+      "loss": 0.4165,
+      "step": 458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.868846836516637e-05,
+      "loss": 0.3933,
+      "step": 460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8675963103909636e-05,
+      "loss": 0.4746,
+      "step": 462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.866340273236926e-05,
+      "loss": 0.4893,
+      "step": 464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8650787330329546e-05,
+      "loss": 0.516,
+      "step": 466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8638116977924346e-05,
+      "loss": 0.4391,
+      "step": 468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.862539175563657e-05,
+      "loss": 0.3998,
+      "step": 470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.861261174429765e-05,
+      "loss": 0.4525,
+      "step": 472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8599777025087068e-05,
+      "loss": 0.4023,
+      "step": 474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.858688767953178e-05,
+      "loss": 0.397,
+      "step": 476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8573943789505762e-05,
+      "loss": 0.4845,
+      "step": 478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8560945437229443e-05,
+      "loss": 0.4518,
+      "step": 480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8547892705269207e-05,
+      "loss": 0.5362,
+      "step": 482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8534785676536856e-05,
+      "loss": 0.4601,
+      "step": 484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8521624434289094e-05,
+      "loss": 0.4814,
+      "step": 486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.850840906212699e-05,
+      "loss": 0.4707,
+      "step": 488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.849513964399545e-05,
+      "loss": 0.4144,
+      "step": 490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8481816264182678e-05,
+      "loss": 0.434,
+      "step": 492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8468439007319663e-05,
+      "loss": 0.4782,
+      "step": 494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8455007958379604e-05,
+      "loss": 0.3848,
+      "step": 496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8441523202677406e-05,
+      "loss": 0.4541,
+      "step": 498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8427984825869114e-05,
+      "loss": 0.4708,
+      "step": 500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8414392913951382e-05,
+      "loss": 0.5103,
+      "step": 502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8400747553260915e-05,
+      "loss": 0.4201,
+      "step": 504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8387048830473948e-05,
+      "loss": 0.4586,
+      "step": 506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8373296832605647e-05,
+      "loss": 0.4667,
+      "step": 508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8359491647009608e-05,
+      "loss": 0.4846,
+      "step": 510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.834563336137727e-05,
+      "loss": 0.5255,
+      "step": 512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8331722063737365e-05,
+      "loss": 0.482,
+      "step": 514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8317757842455363e-05,
+      "loss": 0.4211,
+      "step": 516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.830374078623291e-05,
+      "loss": 0.4852,
+      "step": 518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8289670984107263e-05,
+      "loss": 0.4299,
+      "step": 520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8275548525450722e-05,
+      "loss": 0.5044,
+      "step": 522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8261373499970064e-05,
+      "loss": 0.4072,
+      "step": 524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8247145997705977e-05,
+      "loss": 0.4478,
+      "step": 526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.823286610903248e-05,
+      "loss": 0.4962,
+      "step": 528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8218533924656367e-05,
+      "loss": 0.4658,
+      "step": 530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8204149535616596e-05,
+      "loss": 0.4124,
+      "step": 532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8189713033283755e-05,
+      "loss": 0.4149,
+      "step": 534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.817522450935944e-05,
+      "loss": 0.4327,
+      "step": 536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8160684055875704e-05,
+      "loss": 0.4469,
+      "step": 538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8146091765194458e-05,
+      "loss": 0.44,
+      "step": 540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8131447730006885e-05,
+      "loss": 0.4911,
+      "step": 542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8116752043332848e-05,
+      "loss": 0.4848,
+      "step": 544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.810200479852031e-05,
+      "loss": 0.4297,
+      "step": 546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8087206089244728e-05,
+      "loss": 0.4205,
+      "step": 548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8072356009508473e-05,
+      "loss": 0.3892,
+      "step": 550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.805745465364022e-05,
+      "loss": 0.4519,
+      "step": 552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8042502116294355e-05,
+      "loss": 0.4376,
+      "step": 554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8027498492450367e-05,
+      "loss": 0.4538,
+      "step": 556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8012443877412253e-05,
+      "loss": 0.4672,
+      "step": 558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.799733836680791e-05,
+      "loss": 0.4034,
+      "step": 560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7982182056588536e-05,
+      "loss": 0.4613,
+      "step": 562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.796697504302799e-05,
+      "loss": 0.4664,
+      "step": 564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.795171742272222e-05,
+      "loss": 0.4271,
+      "step": 566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7936409292588627e-05,
+      "loss": 0.4741,
+      "step": 568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.792105074986545e-05,
+      "loss": 0.3175,
+      "step": 570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7905641892111152e-05,
+      "loss": 0.4154,
+      "step": 572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7890182817203806e-05,
+      "loss": 0.4558,
+      "step": 574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7874673623340463e-05,
+      "loss": 0.465,
+      "step": 576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.785911440903653e-05,
+      "loss": 0.4688,
+      "step": 578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7843505273125164e-05,
+      "loss": 0.4411,
+      "step": 580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7827846314756604e-05,
+      "loss": 0.4286,
+      "step": 582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7812137633397577e-05,
+      "loss": 0.4425,
+      "step": 584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7796379328830652e-05,
+      "loss": 0.4126,
+      "step": 586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.778057150115361e-05,
+      "loss": 0.479,
+      "step": 588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.77647142507788e-05,
+      "loss": 0.3994,
+      "step": 590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7748807678432514e-05,
+      "loss": 0.4574,
+      "step": 592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7732851885154336e-05,
+      "loss": 0.3901,
+      "step": 594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7716846972296505e-05,
+      "loss": 0.435,
+      "step": 596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7700793041523272e-05,
+      "loss": 0.4337,
+      "step": 598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7684690194810256e-05,
+      "loss": 0.4196,
+      "step": 600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7668538534443782e-05,
+      "loss": 0.4508,
+      "step": 602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7652338163020257e-05,
+      "loss": 0.4583,
+      "step": 604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.76360891834455e-05,
+      "loss": 0.4499,
+      "step": 606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7619791698934077e-05,
+      "loss": 0.4263,
+      "step": 608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7603445813008685e-05,
+      "loss": 0.3721,
+      "step": 610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7587051629499452e-05,
+      "loss": 0.3788,
+      "step": 612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7570609252543302e-05,
+      "loss": 0.4405,
+      "step": 614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.755411878658329e-05,
+      "loss": 0.4939,
+      "step": 616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7537580336367925e-05,
+      "loss": 0.4188,
+      "step": 618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7520994006950526e-05,
+      "loss": 0.4425,
+      "step": 620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7504359903688537e-05,
+      "loss": 0.37,
+      "step": 622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.748767813224287e-05,
+      "loss": 0.4633,
+      "step": 624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.747094879857722e-05,
+      "loss": 0.3729,
+      "step": 626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7454172008957417e-05,
+      "loss": 0.4312,
+      "step": 628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7437347869950713e-05,
+      "loss": 0.4367,
+      "step": 630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7420476488425138e-05,
+      "loss": 0.4091,
+      "step": 632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.740355797154881e-05,
+      "loss": 0.469,
+      "step": 634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7386592426789252e-05,
+      "loss": 0.4872,
+      "step": 636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7369579961912712e-05,
+      "loss": 0.4932,
+      "step": 638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7352520684983474e-05,
+      "loss": 0.3848,
+      "step": 640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7335414704363178e-05,
+      "loss": 0.3694,
+      "step": 642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7318262128710132e-05,
+      "loss": 0.5099,
+      "step": 644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7301063066978617e-05,
+      "loss": 0.4407,
+      "step": 646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.728381762841819e-05,
+      "loss": 0.4409,
+      "step": 648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7266525922573e-05,
+      "loss": 0.4444,
+      "step": 650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.72491880592811e-05,
+      "loss": 0.4079,
+      "step": 652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7231804148673717e-05,
+      "loss": 0.4502,
+      "step": 654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7214374301174594e-05,
+      "loss": 0.49,
+      "step": 656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.719689862749926e-05,
+      "loss": 0.4778,
+      "step": 658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7179377238654325e-05,
+      "loss": 0.3734,
+      "step": 660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.716181024593681e-05,
+      "loss": 0.4956,
+      "step": 662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.714419776093338e-05,
+      "loss": 0.3712,
+      "step": 664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7126539895519698e-05,
+      "loss": 0.3779,
+      "step": 666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.710883676185968e-05,
+      "loss": 0.4457,
+      "step": 668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.709108847240478e-05,
+      "loss": 0.4161,
+      "step": 670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7073295139893296e-05,
+      "loss": 0.4459,
+      "step": 672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.705545687734963e-05,
+      "loss": 0.4465,
+      "step": 674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7037573798083598e-05,
+      "loss": 0.4284,
+      "step": 676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.701964601568968e-05,
+      "loss": 0.3806,
+      "step": 678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7001673644046322e-05,
+      "loss": 0.4591,
+      "step": 680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6983656797315197e-05,
+      "loss": 0.4809,
+      "step": 682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6965595589940496e-05,
+      "loss": 0.3811,
+      "step": 684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947490136648182e-05,
+      "loss": 0.4223,
+      "step": 686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6929340552445283e-05,
+      "loss": 0.4698,
+      "step": 688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6911146952619132e-05,
+      "loss": 0.4059,
+      "step": 690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689290945273667e-05,
+      "loss": 0.4451,
+      "step": 692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6874628168643683e-05,
+      "loss": 0.3428,
+      "step": 694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.685630321646408e-05,
+      "loss": 0.4552,
+      "step": 696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.683793471259915e-05,
+      "loss": 0.6389,
+      "step": 698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.681952277372683e-05,
+      "loss": 0.4356,
+      "step": 700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.680106751680096e-05,
+      "loss": 0.4199,
+      "step": 702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6782569059050535e-05,
+      "loss": 0.4369,
+      "step": 704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.676402751797896e-05,
+      "loss": 0.3924,
+      "step": 706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.674544301136332e-05,
+      "loss": 0.386,
+      "step": 708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.672681565725361e-05,
+      "loss": 0.464,
+      "step": 710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6708145573972005e-05,
+      "loss": 0.4597,
+      "step": 712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6689432880112078e-05,
+      "loss": 0.4164,
+      "step": 714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6670677694538096e-05,
+      "loss": 0.3761,
+      "step": 716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6651880136384215e-05,
+      "loss": 0.4499,
+      "step": 718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6633040325053746e-05,
+      "loss": 0.438,
+      "step": 720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.661415838021841e-05,
+      "loss": 0.4526,
+      "step": 722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.659523442181754e-05,
+      "loss": 0.4427,
+      "step": 724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6576268570057363e-05,
+      "loss": 0.5268,
+      "step": 726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.655726094541021e-05,
+      "loss": 0.4135,
+      "step": 728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.653821166861374e-05,
+      "loss": 0.42,
+      "step": 730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6519120860670215e-05,
+      "loss": 0.449,
+      "step": 732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6499988642845686e-05,
+      "loss": 0.4751,
+      "step": 734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6480815136669248e-05,
+      "loss": 0.3826,
+      "step": 736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6461600463932266e-05,
+      "loss": 0.4712,
+      "step": 738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6442344746687594e-05,
+      "loss": 0.4128,
+      "step": 740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.64230481072488e-05,
+      "loss": 0.4679,
+      "step": 742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.640371066818941e-05,
+      "loss": 0.4768,
+      "step": 744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.638433255234208e-05,
+      "loss": 0.4785,
+      "step": 746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6364913882797875e-05,
+      "loss": 0.4334,
+      "step": 748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6345454782905454e-05,
+      "loss": 0.4015,
+      "step": 750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6325955376270286e-05,
+      "loss": 0.439,
+      "step": 752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.630641578675387e-05,
+      "loss": 0.3951,
+      "step": 754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6296630962191733e-05,
+      "loss": 0.5453,
+      "step": 756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.62770313311519e-05,
+      "loss": 0.3902,
+      "step": 758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.625739182799955e-05,
+      "loss": 0.3943,
+      "step": 760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6237712577486092e-05,
+      "loss": 0.3312,
+      "step": 762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.62179937046154e-05,
+      "loss": 0.4366,
+      "step": 764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6198235334643045e-05,
+      "loss": 0.3924,
+      "step": 766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6178437593075487e-05,
+      "loss": 0.378,
+      "step": 768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6158600605669264e-05,
+      "loss": 0.4624,
+      "step": 770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.613872449843022e-05,
+      "loss": 0.4411,
+      "step": 772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6118809397612678e-05,
+      "loss": 0.4695,
+      "step": 774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6098855429718662e-05,
+      "loss": 0.4348,
+      "step": 776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.607886272149708e-05,
+      "loss": 0.4048,
+      "step": 778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6058831399942917e-05,
+      "loss": 0.3485,
+      "step": 780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6038761592296435e-05,
+      "loss": 0.4146,
+      "step": 782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6018653426042357e-05,
+      "loss": 0.4398,
+      "step": 784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5998507028909074e-05,
+      "loss": 0.5815,
+      "step": 786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.597832252886781e-05,
+      "loss": 0.4502,
+      "step": 788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5958100054131828e-05,
+      "loss": 0.4275,
+      "step": 790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5937839733155603e-05,
+      "loss": 0.4269,
+      "step": 792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.591754169463402e-05,
+      "loss": 0.4211,
+      "step": 794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5897206067501544e-05,
+      "loss": 0.4194,
+      "step": 796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5876832980931405e-05,
+      "loss": 0.3833,
+      "step": 798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5856422564334772e-05,
+      "loss": 0.4176,
+      "step": 800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5835974947359952e-05,
+      "loss": 0.5327,
+      "step": 802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.581549025989154e-05,
+      "loss": 0.4776,
+      "step": 804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5794968632049598e-05,
+      "loss": 0.3573,
+      "step": 806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5774410194188856e-05,
+      "loss": 0.464,
+      "step": 808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5753815076897848e-05,
+      "loss": 0.4549,
+      "step": 810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57331834109981e-05,
+      "loss": 0.463,
+      "step": 812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5712515327543307e-05,
+      "loss": 0.4438,
+      "step": 814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5691810957818475e-05,
+      "loss": 0.4306,
+      "step": 816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5671070433339116e-05,
+      "loss": 0.4135,
+      "step": 818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5650293885850393e-05,
+      "loss": 0.4706,
+      "step": 820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5629481447326297e-05,
+      "loss": 0.4427,
+      "step": 822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5608633249968783e-05,
+      "loss": 0.4661,
+      "step": 824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.558774942620697e-05,
+      "loss": 0.3674,
+      "step": 826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5566830108696265e-05,
+      "loss": 0.4204,
+      "step": 828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5545875430317546e-05,
+      "loss": 0.4685,
+      "step": 830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5524885524176287e-05,
+      "loss": 0.4583,
+      "step": 832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.550386052360174e-05,
+      "loss": 0.4306,
+      "step": 834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.548280056214609e-05,
+      "loss": 0.4203,
+      "step": 836
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.546170577358358e-05,
+      "loss": 0.4664,
+      "step": 838
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.544057629190969e-05,
+      "loss": 0.3553,
+      "step": 840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.541941225134025e-05,
+      "loss": 0.368,
+      "step": 842
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5398213786310643e-05,
+      "loss": 0.3176,
+      "step": 844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.537698103147489e-05,
+      "loss": 0.2801,
+      "step": 846
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5355714121704846e-05,
+      "loss": 0.2576,
+      "step": 848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.53344131920893e-05,
+      "loss": 0.2689,
+      "step": 850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.531307837793315e-05,
+      "loss": 0.3045,
+      "step": 852
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.529170981475653e-05,
+      "loss": 0.2506,
+      "step": 854
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5270307638293943e-05,
+      "loss": 0.2546,
+      "step": 856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.524887198449341e-05,
+      "loss": 0.2853,
+      "step": 858
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5227402989515607e-05,
+      "loss": 0.2772,
+      "step": 860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5205900789732986e-05,
+      "loss": 0.2763,
+      "step": 862
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5184365521728928e-05,
+      "loss": 0.2578,
+      "step": 864
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5162797322296855e-05,
+      "loss": 0.3121,
+      "step": 866
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5141196328439377e-05,
+      "loss": 0.3037,
+      "step": 868
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5119562677367421e-05,
+      "loss": 0.2877,
+      "step": 870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5097896506499349e-05,
+      "loss": 0.2856,
+      "step": 872
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5076197953460087e-05,
+      "loss": 0.3417,
+      "step": 874
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5054467156080262e-05,
+      "loss": 0.285,
+      "step": 876
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5032704252395315e-05,
+      "loss": 0.3137,
+      "step": 878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5010909380644636e-05,
+      "loss": 0.2204,
+      "step": 880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4989082679270668e-05,
+      "loss": 0.2808,
+      "step": 882
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.496722428691804e-05,
+      "loss": 0.2691,
+      "step": 884
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4945334342432688e-05,
+      "loss": 0.2638,
+      "step": 886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.492341298486097e-05,
+      "loss": 0.26,
+      "step": 888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.490146035344878e-05,
+      "loss": 0.2764,
+      "step": 890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4879476587640657e-05,
+      "loss": 0.2558,
+      "step": 892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4868473072968645e-05,
+      "loss": 0.5349,
+      "step": 894
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4846442867457533e-05,
+      "loss": 0.2937,
+      "step": 896
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4824381877025154e-05,
+      "loss": 0.2684,
+      "step": 898
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4802290241804355e-05,
+      "loss": 0.2491,
+      "step": 900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.478016810212265e-05,
+      "loss": 0.2634,
+      "step": 902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4758015598501308e-05,
+      "loss": 0.2889,
+      "step": 904
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.473583287165448e-05,
+      "loss": 0.2843,
+      "step": 906
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4713620062488296e-05,
+      "loss": 0.2705,
+      "step": 908
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4691377312099965e-05,
+      "loss": 0.2765,
+      "step": 910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4669104761776892e-05,
+      "loss": 0.2595,
+      "step": 912
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4646802552995767e-05,
+      "loss": 0.2101,
+      "step": 914
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4624470827421675e-05,
+      "loss": 0.263,
+      "step": 916
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4602109726907197e-05,
+      "loss": 0.2592,
+      "step": 918
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4579719393491496e-05,
+      "loss": 0.2732,
+      "step": 920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.455729996939944e-05,
+      "loss": 0.3056,
+      "step": 922
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4534851597040666e-05,
+      "loss": 0.2886,
+      "step": 924
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.45123744190087e-05,
+      "loss": 0.2493,
+      "step": 926
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4489868578080046e-05,
+      "loss": 0.271,
+      "step": 928
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4467334217213274e-05,
+      "loss": 0.2752,
+      "step": 930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4444771479548115e-05,
+      "loss": 0.3108,
+      "step": 932
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4422180508404544e-05,
+      "loss": 0.2946,
+      "step": 934
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.439956144728189e-05,
+      "loss": 0.2401,
+      "step": 936
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4376914439857905e-05,
+      "loss": 0.3501,
+      "step": 938
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4354239629987857e-05,
+      "loss": 0.2895,
+      "step": 940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4331537161703612e-05,
+      "loss": 0.2632,
+      "step": 942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4308807179212736e-05,
+      "loss": 0.261,
+      "step": 944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4286049826897559e-05,
+      "loss": 0.3207,
+      "step": 946
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4263265249314269e-05,
+      "loss": 0.2592,
+      "step": 948
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4240453591191984e-05,
+      "loss": 0.2468,
+      "step": 950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4217614997431847e-05,
+      "loss": 0.2483,
+      "step": 952
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.41947496131061e-05,
+      "loss": 0.2657,
+      "step": 954
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4171857583457154e-05,
+      "loss": 0.2389,
+      "step": 956
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4148939053896669e-05,
+      "loss": 0.2404,
+      "step": 958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4125994170004644e-05,
+      "loss": 0.2539,
+      "step": 960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4103023077528482e-05,
+      "loss": 0.2721,
+      "step": 962
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4080025922382056e-05,
+      "loss": 0.3314,
+      "step": 964
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4057002850644796e-05,
+      "loss": 0.2668,
+      "step": 966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4033954008560758e-05,
+      "loss": 0.2295,
+      "step": 968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.401087954253769e-05,
+      "loss": 0.284,
+      "step": 970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3987779599146105e-05,
+      "loss": 0.2595,
+      "step": 972
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.396465432511835e-05,
+      "loss": 0.2849,
+      "step": 974
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3941503867347672e-05,
+      "loss": 0.271,
+      "step": 976
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3918328372887295e-05,
+      "loss": 0.2943,
+      "step": 978
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895127988949471e-05,
+      "loss": 0.2751,
+      "step": 980
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3871902862904544e-05,
+      "loss": 0.276,
+      "step": 982
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3848653142280037e-05,
+      "loss": 0.2251,
+      "step": 984
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3825378974759696e-05,
+      "loss": 0.2722,
+      "step": 986
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3802080508182543e-05,
+      "loss": 0.2927,
+      "step": 988
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.377875789054196e-05,
+      "loss": 0.2473,
+      "step": 990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.376708757136279e-05,
+      "loss": 0.3166,
+      "step": 992
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3743729004949972e-05,
+      "loss": 0.3079,
+      "step": 994
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3720346658126286e-05,
+      "loss": 0.2695,
+      "step": 996
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696940679417918e-05,
+      "loss": 0.3125,
+      "step": 998
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3673511217501172e-05,
+      "loss": 0.2874,
+      "step": 1000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3650058421201517e-05,
+      "loss": 0.31,
+      "step": 1002
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.362658243949265e-05,
+      "loss": 0.2795,
+      "step": 1004
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3603083421495535e-05,
+      "loss": 0.2693,
+      "step": 1006
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3579561516477467e-05,
+      "loss": 0.2659,
+      "step": 1008
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.355601687385112e-05,
+      "loss": 0.2909,
+      "step": 1010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3532449643173604e-05,
+      "loss": 0.262,
+      "step": 1012
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3508859974145504e-05,
+      "loss": 0.2538,
+      "step": 1014
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3485248016609937e-05,
+      "loss": 0.2674,
+      "step": 1016
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3461613920551598e-05,
+      "loss": 0.2863,
+      "step": 1018
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3437957836095804e-05,
+      "loss": 0.3213,
+      "step": 1020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414279913507548e-05,
+      "loss": 0.2932,
+      "step": 1022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3390580303190541e-05,
+      "loss": 0.2604,
+      "step": 1024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3366859155686253e-05,
+      "loss": 0.275,
+      "step": 1026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3343116621672959e-05,
+      "loss": 0.2625,
+      "step": 1028
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3319352851964787e-05,
+      "loss": 0.2664,
+      "step": 1030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3295567997510747e-05,
+      "loss": 0.2567,
+      "step": 1032
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3271762209393793e-05,
+      "loss": 0.2469,
+      "step": 1034
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3247935638829838e-05,
+      "loss": 0.2596,
+      "step": 1036
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3224088437166818e-05,
+      "loss": 0.3033,
+      "step": 1038
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320022075588371e-05,
+      "loss": 0.3582,
+      "step": 1040
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3176332746589587e-05,
+      "loss": 0.2339,
+      "step": 1042
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3152424561022634e-05,
+      "loss": 0.2622,
+      "step": 1044
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3128496351049216e-05,
+      "loss": 0.2388,
+      "step": 1046
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3104548268662873e-05,
+      "loss": 0.2322,
+      "step": 1048
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3080580465983397e-05,
+      "loss": 0.3108,
+      "step": 1050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3056593095255825e-05,
+      "loss": 0.2339,
+      "step": 1052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3032586308849512e-05,
+      "loss": 0.2731,
+      "step": 1054
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3008560259257117e-05,
+      "loss": 0.2677,
+      "step": 1056
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2984515099093687e-05,
+      "loss": 0.2907,
+      "step": 1058
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2960450981095643e-05,
+      "loss": 0.2836,
+      "step": 1060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2936368058119828e-05,
+      "loss": 0.2621,
+      "step": 1062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912266483142545e-05,
+      "loss": 0.3009,
+      "step": 1064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888146409258575e-05,
+      "loss": 0.252,
+      "step": 1066
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2864007989680194e-05,
+      "loss": 0.3354,
+      "step": 1068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2839851377736216e-05,
+      "loss": 0.2908,
+      "step": 1070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.281567672687102e-05,
+      "loss": 0.284,
+      "step": 1072
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791484190643571e-05,
+      "loss": 0.2882,
+      "step": 1074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2767273922726427e-05,
+      "loss": 0.3096,
+      "step": 1076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2743046076904795e-05,
+      "loss": 0.2674,
+      "step": 1078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.271880080707553e-05,
+      "loss": 0.2207,
+      "step": 1080
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2694538267246168e-05,
+      "loss": 0.2605,
+      "step": 1082
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2670258611533947e-05,
+      "loss": 0.2598,
+      "step": 1084
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2645961994164822e-05,
+      "loss": 0.2856,
+      "step": 1086
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2621648569472491e-05,
+      "loss": 0.2436,
+      "step": 1088
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597318491897416e-05,
+      "loss": 0.2606,
+      "step": 1090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257297191598584e-05,
+      "loss": 0.2602,
+      "step": 1092
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2548608996388792e-05,
+      "loss": 0.2465,
+      "step": 1094
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2524229887861132e-05,
+      "loss": 0.2536,
+      "step": 1096
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499834745260553e-05,
+      "loss": 0.2859,
+      "step": 1098
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475423723546584e-05,
+      "loss": 0.2539,
+      "step": 1100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.245099697777963e-05,
+      "loss": 0.2299,
+      "step": 1102
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2426554663119975e-05,
+      "loss": 0.4215,
+      "step": 1104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402096934826794e-05,
+      "loss": 0.2575,
+      "step": 1106
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237762394825718e-05,
+      "loss": 0.3257,
+      "step": 1108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2353135858865128e-05,
+      "loss": 0.2778,
+      "step": 1110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.232863282220059e-05,
+      "loss": 0.2592,
+      "step": 1112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.230411499390845e-05,
+      "loss": 0.2539,
+      "step": 1114
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279582529727552e-05,
+      "loss": 0.2831,
+      "step": 1116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2255035585489705e-05,
+      "loss": 0.2806,
+      "step": 1118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2230474317118708e-05,
+      "loss": 0.2777,
+      "step": 1120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2205898880629336e-05,
+      "loss": 0.3334,
+      "step": 1122
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2181309432126366e-05,
+      "loss": 0.302,
+      "step": 1124
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156706127803578e-05,
+      "loss": 0.2659,
+      "step": 1126
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2132089123942764e-05,
+      "loss": 0.297,
+      "step": 1128
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2107458576912743e-05,
+      "loss": 0.3207,
+      "step": 1130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082814643168357e-05,
+      "loss": 0.2224,
+      "step": 1132
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058157479249475e-05,
+      "loss": 0.295,
+      "step": 1134
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2033487241780014e-05,
+      "loss": 0.2238,
+      "step": 1136
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2008804087466931e-05,
+      "loss": 0.277,
+      "step": 1138
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1984108173099238e-05,
+      "loss": 0.2906,
+      "step": 1140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959399655546989e-05,
+      "loss": 0.2649,
+      "step": 1142
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934678691760296e-05,
+      "loss": 0.3147,
+      "step": 1144
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.190994543876834e-05,
+      "loss": 0.2761,
+      "step": 1146
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188520005367836e-05,
+      "loss": 0.2252,
+      "step": 1148
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860442693674648e-05,
+      "loss": 0.2521,
+      "step": 1150
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1835673516017571e-05,
+      "loss": 0.2618,
+      "step": 1152
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1810892678042565e-05,
+      "loss": 0.2869,
+      "step": 1154
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1786100337159132e-05,
+      "loss": 0.2124,
+      "step": 1156
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177369990233723e-05,
+      "loss": 0.3826,
+      "step": 1158
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748890602393521e-05,
+      "loss": 0.2805,
+      "step": 1160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172407019338261e-05,
+      "loss": 0.2467,
+      "step": 1162
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1699238832965358e-05,
+      "loss": 0.2405,
+      "step": 1164
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674396678872186e-05,
+      "loss": 0.3017,
+      "step": 1166
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164954388890207e-05,
+      "loss": 0.2584,
+      "step": 1168
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162468062092156e-05,
+      "loss": 0.3141,
+      "step": 1170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1599807032863756e-05,
+      "loss": 0.3254,
+      "step": 1172
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574923282727314e-05,
+      "loss": 0.2703,
+      "step": 1174
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550029528575428e-05,
+      "loss": 0.2207,
+      "step": 1176
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152512592853486e-05,
+      "loss": 0.2634,
+      "step": 1178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1500212640794895e-05,
+      "loss": 0.3368,
+      "step": 1180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1475289823606364e-05,
+      "loss": 0.2535,
+      "step": 1182
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450357635280628e-05,
+      "loss": 0.287,
+      "step": 1184
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1425416234188578e-05,
+      "loss": 0.3052,
+      "step": 1186
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1400465778759611e-05,
+      "loss": 0.2909,
+      "step": 1188
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1375506427480658e-05,
+      "loss": 0.2904,
+      "step": 1190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135053833889514e-05,
+      "loss": 0.339,
+      "step": 1192
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325561671601987e-05,
+      "loss": 0.292,
+      "step": 1194
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1300576584254617e-05,
+      "loss": 0.2424,
+      "step": 1196
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.127558323555994e-05,
+      "loss": 0.267,
+      "step": 1198
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125058178427733e-05,
+      "loss": 0.2641,
+      "step": 1200
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1225572389217643e-05,
+      "loss": 0.269,
+      "step": 1202
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200555209242182e-05,
+      "loss": 0.2903,
+      "step": 1204
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1175530403261716e-05,
+      "loss": 0.2622,
+      "step": 1206
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1150498130235435e-05,
+      "loss": 0.3199,
+      "step": 1208
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125458549169977e-05,
+      "loss": 0.2469,
+      "step": 1210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1100411819118387e-05,
+      "loss": 0.2781,
+      "step": 1212
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1075358099179136e-05,
+      "loss": 0.3293,
+      "step": 1214
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050297548495084e-05,
+      "loss": 0.2065,
+      "step": 1216
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1025230326252484e-05,
+      "loss": 0.2548,
+      "step": 1218
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1000156591679971e-05,
+      "loss": 0.3063,
+      "step": 1220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975076504047535e-05,
+      "loss": 0.3099,
+      "step": 1222
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0949990222665532e-05,
+      "loss": 0.2805,
+      "step": 1224
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924897906883663e-05,
+      "loss": 0.288,
+      "step": 1226
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899799716089949e-05,
+      "loss": 0.3014,
+      "step": 1228
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874695809709737e-05,
+      "loss": 0.2768,
+      "step": 1230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0849586347204677e-05,
+      "loss": 0.2894,
+      "step": 1232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824471488071714e-05,
+      "loss": 0.2718,
+      "step": 1234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799351391842074e-05,
+      "loss": 0.2476,
+      "step": 1236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774226218080244e-05,
+      "loss": 0.2318,
+      "step": 1238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0749096126382965e-05,
+      "loss": 0.2545,
+      "step": 1240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723961276378225e-05,
+      "loss": 0.2708,
+      "step": 1242
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0698821827724225e-05,
+      "loss": 0.3471,
+      "step": 1244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673677940108386e-05,
+      "loss": 0.2528,
+      "step": 1246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0648529773246324e-05,
+      "loss": 0.2625,
+      "step": 1248
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623377486880831e-05,
+      "loss": 0.2634,
+      "step": 1250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598221240780874e-05,
+      "loss": 0.2506,
+      "step": 1252
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0573061194740568e-05,
+      "loss": 0.2659,
+      "step": 1254
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054789750857817e-05,
+      "loss": 0.239,
+      "step": 1256
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052273034213505e-05,
+      "loss": 0.2465,
+      "step": 1258
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0497559855274699e-05,
+      "loss": 0.2512,
+      "step": 1260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472386207881684e-05,
+      "loss": 0.303,
+      "step": 1262
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0447209559860658e-05,
+      "loss": 0.2542,
+      "step": 1264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422030071135336e-05,
+      "loss": 0.2995,
+      "step": 1266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0396847901647469e-05,
+      "loss": 0.2597,
+      "step": 1268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037166321135584e-05,
+      "loss": 0.2773,
+      "step": 1270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0346476160235246e-05,
+      "loss": 0.2771,
+      "step": 1272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0321286908275476e-05,
+      "loss": 0.2906,
+      "step": 1274
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296095615480309e-05,
+      "loss": 0.3002,
+      "step": 1276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270902441866474e-05,
+      "loss": 0.2267,
+      "step": 1278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0245707547462654e-05,
+      "loss": 0.2545,
+      "step": 1280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0220511092308463e-05,
+      "loss": 0.2412,
+      "step": 1282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0195313236453431e-05,
+      "loss": 0.2567,
+      "step": 1284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0170114139955975e-05,
+      "loss": 0.2589,
+      "step": 1286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144913962882406e-05,
+      "loss": 0.2834,
+      "step": 1288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119712865305891e-05,
+      "loss": 0.2504,
+      "step": 1290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094511007305445e-05,
+      "loss": 0.2788,
+      "step": 1292
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069308548964915e-05,
+      "loss": 0.2664,
+      "step": 1294
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044105650371961e-05,
+      "loss": 0.2695,
+      "step": 1296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0018902471617037e-05,
+      "loss": 0.2309,
+      "step": 1298
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993699172792381e-06,
+      "loss": 0.2949,
+      "step": 1300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96849591399099e-06,
+      "loss": 0.3248,
+      "step": 1302
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.943292855305611e-06,
+      "loss": 0.2719,
+      "step": 1304
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.918090156827712e-06,
+      "loss": 0.2417,
+      "step": 1306
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.892887978646483e-06,
+      "loss": 0.2371,
+      "step": 1308
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.867686480847801e-06,
+      "loss": 0.271,
+      "step": 1310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842485823513222e-06,
+      "loss": 0.2791,
+      "step": 1312
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.817286166718971e-06,
+      "loss": 0.2759,
+      "step": 1314
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.792087670534908e-06,
+      "loss": 0.2867,
+      "step": 1316
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.766890495023522e-06,
+      "loss": 0.2964,
+      "step": 1318
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.741694800238923e-06,
+      "loss": 0.2466,
+      "step": 1320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.716500746225802e-06,
+      "loss": 0.2745,
+      "step": 1322
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691308493018439e-06,
+      "loss": 0.2429,
+      "step": 1324
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666118200639667e-06,
+      "loss": 0.2561,
+      "step": 1326
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.640930029099863e-06,
+      "loss": 0.2462,
+      "step": 1328
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615744138395941e-06,
+      "loss": 0.2294,
+      "step": 1330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590560688510323e-06,
+      "loss": 0.2462,
+      "step": 1332
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.565379839409916e-06,
+      "loss": 0.2755,
+      "step": 1334
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.540201751045127e-06,
+      "loss": 0.2623,
+      "step": 1336
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.515026583348811e-06,
+      "loss": 0.3047,
+      "step": 1338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.489854496235278e-06,
+      "loss": 0.2489,
+      "step": 1340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.464685649599266e-06,
+      "loss": 0.23,
+      "step": 1342
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439520203314927e-06,
+      "loss": 0.2517,
+      "step": 1344
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.414358317234826e-06,
+      "loss": 0.3041,
+      "step": 1346
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3892001511889e-06,
+      "loss": 0.2922,
+      "step": 1348
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.364045864983454e-06,
+      "loss": 0.2451,
+      "step": 1350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338895618400168e-06,
+      "loss": 0.2457,
+      "step": 1352
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.313749571195041e-06,
+      "loss": 0.2488,
+      "step": 1354
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28860788309741e-06,
+      "loss": 0.2599,
+      "step": 1356
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263470713808917e-06,
+      "loss": 0.2192,
+      "step": 1358
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238338223002496e-06,
+      "loss": 0.2296,
+      "step": 1360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213210570321374e-06,
+      "loss": 0.2907,
+      "step": 1362
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188087915378037e-06,
+      "loss": 0.2644,
+      "step": 1364
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162970417753229e-06,
+      "loss": 0.2307,
+      "step": 1366
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137858236994932e-06,
+      "loss": 0.2493,
+      "step": 1368
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112751532617361e-06,
+      "loss": 0.2546,
+      "step": 1370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087650464099937e-06,
+      "loss": 0.2835,
+      "step": 1372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062555190886287e-06,
+      "loss": 0.2701,
+      "step": 1374
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.037465872383219e-06,
+      "loss": 0.2514,
+      "step": 1376
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.012382667959724e-06,
+      "loss": 0.2423,
+      "step": 1378
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987305736945955e-06,
+      "loss": 0.2289,
+      "step": 1380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962235238632208e-06,
+      "loss": 0.2504,
+      "step": 1382
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.937171332267927e-06,
+      "loss": 0.2912,
+      "step": 1384
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912114177060681e-06,
+      "loss": 0.2368,
+      "step": 1386
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887063932175156e-06,
+      "loss": 0.2823,
+      "step": 1388
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.862020756732141e-06,
+      "loss": 0.2289,
+      "step": 1390
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.836984809807514e-06,
+      "loss": 0.2332,
+      "step": 1392
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811956250431253e-06,
+      "loss": 0.2627,
+      "step": 1394
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.786935237586394e-06,
+      "loss": 0.2613,
+      "step": 1396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761921930208044e-06,
+      "loss": 0.2353,
+      "step": 1398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73691648718236e-06,
+      "loss": 0.2668,
+      "step": 1400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.71191906734555e-06,
+      "loss": 0.2292,
+      "step": 1402
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686929829482862e-06,
+      "loss": 0.2577,
+      "step": 1404
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661948932327558e-06,
+      "loss": 0.2177,
+      "step": 1406
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636976534559926e-06,
+      "loss": 0.2849,
+      "step": 1408
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.61201279480627e-06,
+      "loss": 0.2575,
+      "step": 1410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.587057871637891e-06,
+      "loss": 0.267,
+      "step": 1412
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.562111923570091e-06,
+      "loss": 0.2572,
+      "step": 1414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.537175109061154e-06,
+      "loss": 0.2546,
+      "step": 1416
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.512247586511354e-06,
+      "loss": 0.3069,
+      "step": 1418
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.487329514261948e-06,
+      "loss": 0.2915,
+      "step": 1420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46242105059415e-06,
+      "loss": 0.3033,
+      "step": 1422
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.437522353728147e-06,
+      "loss": 0.2389,
+      "step": 1424
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.412633581822086e-06,
+      "loss": 0.4257,
+      "step": 1426
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387754892971073e-06,
+      "loss": 0.2206,
+      "step": 1428
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.36288644520616e-06,
+      "loss": 0.2217,
+      "step": 1430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.338028396493345e-06,
+      "loss": 0.27,
+      "step": 1432
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.313180904732578e-06,
+      "loss": 0.266,
+      "step": 1434
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.288344127756755e-06,
+      "loss": 0.2922,
+      "step": 1436
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.2122,
+      "step": 1438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.238703349150169e-06,
+      "loss": 0.2822,
+      "step": 1440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.213899662840871e-06,
+      "loss": 0.2687,
+      "step": 1442
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.189107321957437e-06,
+      "loss": 0.2783,
+      "step": 1444
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.164326483982434e-06,
+      "loss": 0.2499,
+      "step": 1446
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139557306325359e-06,
+      "loss": 0.2408,
+      "step": 1448
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.114799946321647e-06,
+      "loss": 0.2584,
+      "step": 1450
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.090054561231659e-06,
+      "loss": 0.237,
+      "step": 1452
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065321308239706e-06,
+      "loss": 0.2553,
+      "step": 1454
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.040600344453013e-06,
+      "loss": 0.2224,
+      "step": 1456
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.015891826900764e-06,
+      "loss": 0.3115,
+      "step": 1458
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.99119591253307e-06,
+      "loss": 0.223,
+      "step": 1460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.966512758219991e-06,
+      "loss": 0.2604,
+      "step": 1462
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941842520750529e-06,
+      "loss": 0.262,
+      "step": 1464
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91718535683165e-06,
+      "loss": 0.2583,
+      "step": 1466
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.892541423087258e-06,
+      "loss": 0.2318,
+      "step": 1468
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.867910876057238e-06,
+      "loss": 0.2489,
+      "step": 1470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843293872196425e-06,
+      "loss": 0.2609,
+      "step": 1472
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818690567873637e-06,
+      "loss": 0.2592,
+      "step": 1474
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.794101119370668e-06,
+      "loss": 0.269,
+      "step": 1476
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.769525682881295e-06,
+      "loss": 0.2532,
+      "step": 1478
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.744964414510297e-06,
+      "loss": 0.2223,
+      "step": 1480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.720417470272455e-06,
+      "loss": 0.2525,
+      "step": 1482
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695885006091552e-06,
+      "loss": 0.2701,
+      "step": 1484
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.67136717779941e-06,
+      "loss": 0.3059,
+      "step": 1486
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646864141134874e-06,
+      "loss": 0.2591,
+      "step": 1488
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622376051742824e-06,
+      "loss": 0.2618,
+      "step": 1490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5979030651732065e-06,
+      "loss": 0.2814,
+      "step": 1492
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.573445336880029e-06,
+      "loss": 0.2389,
+      "step": 1494
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.549003022220374e-06,
+      "loss": 0.2347,
+      "step": 1496
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.524576276453422e-06,
+      "loss": 0.2642,
+      "step": 1498
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500165254739453e-06,
+      "loss": 0.2591,
+      "step": 1500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.475770112138867e-06,
+      "loss": 0.2591,
+      "step": 1502
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4513910036112105e-06,
+      "loss": 0.2878,
+      "step": 1504
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.427028084014163e-06,
+      "loss": 0.305,
+      "step": 1506
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402681508102585e-06,
+      "loss": 0.263,
+      "step": 1508
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.378351430527511e-06,
+      "loss": 0.2868,
+      "step": 1510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.35403800583518e-06,
+      "loss": 0.2299,
+      "step": 1512
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.329741388466056e-06,
+      "loss": 0.2942,
+      "step": 1514
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.305461732753836e-06,
+      "loss": 0.2993,
+      "step": 1516
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.281199192924473e-06,
+      "loss": 0.263,
+      "step": 1518
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.256953923095209e-06,
+      "loss": 0.2591,
+      "step": 1520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.232726077273575e-06,
+      "loss": 0.2612,
+      "step": 1522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208515809356434e-06,
+      "loss": 0.2354,
+      "step": 1524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.184323273128981e-06,
+      "loss": 0.2504,
+      "step": 1526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.160148622263786e-06,
+      "loss": 0.2437,
+      "step": 1528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.135992010319812e-06,
+      "loss": 0.2154,
+      "step": 1530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123920516899151e-06,
+      "loss": 0.2963,
+      "step": 1532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0997912510091335e-06,
+      "loss": 0.2307,
+      "step": 1534
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.075680407434289e-06,
+      "loss": 0.2826,
+      "step": 1536
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.051588139328276e-06,
+      "loss": 0.2344,
+      "step": 1538
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0275145997267544e-06,
+      "loss": 0.3482,
+      "step": 1540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0034599415464135e-06,
+      "loss": 0.2818,
+      "step": 1542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.979424317584014e-06,
+      "loss": 0.3444,
+      "step": 1544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.955407880515404e-06,
+      "loss": 0.2712,
+      "step": 1546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.931410782894563e-06,
+      "loss": 0.2794,
+      "step": 1548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907433177152618e-06,
+      "loss": 0.2701,
+      "step": 1550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.883475215596882e-06,
+      "loss": 0.2943,
+      "step": 1552
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.859537050409895e-06,
+      "loss": 0.2488,
+      "step": 1554
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.835618833648443e-06,
+      "loss": 0.2451,
+      "step": 1556
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8117207172425996e-06,
+      "loss": 0.3123,
+      "step": 1558
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.787842852994757e-06,
+      "loss": 0.2522,
+      "step": 1560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.763985392578667e-06,
+      "loss": 0.2766,
+      "step": 1562
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.740148487538476e-06,
+      "loss": 0.2473,
+      "step": 1564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.716332289287759e-06,
+      "loss": 0.214,
+      "step": 1566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692536949108562e-06,
+      "loss": 0.2742,
+      "step": 1568
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6687626181504315e-06,
+      "loss": 0.2348,
+      "step": 1570
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.64500944742948e-06,
+      "loss": 0.2363,
+      "step": 1572
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6212775878273925e-06,
+      "loss": 0.2686,
+      "step": 1574
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.59756719009049e-06,
+      "loss": 0.2852,
+      "step": 1576
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5738784048287615e-06,
+      "loss": 0.2272,
+      "step": 1578
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.550211382514922e-06,
+      "loss": 0.2975,
+      "step": 1580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.526566273483439e-06,
+      "loss": 0.2563,
+      "step": 1582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.502943227929586e-06,
+      "loss": 0.2971,
+      "step": 1584
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.479342395908487e-06,
+      "loss": 0.2601,
+      "step": 1586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.455763927334177e-06,
+      "loss": 0.258,
+      "step": 1588
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.432207971978619e-06,
+      "loss": 0.2953,
+      "step": 1590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4086746794707795e-06,
+      "loss": 0.2961,
+      "step": 1592
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.385164199295666e-06,
+      "loss": 0.262,
+      "step": 1594
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3616766807933875e-06,
+      "loss": 0.2151,
+      "step": 1596
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.338212273158188e-06,
+      "loss": 0.2525,
+      "step": 1598
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.314771125437517e-06,
+      "loss": 0.2393,
+      "step": 1600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291353386531074e-06,
+      "loss": 0.2758,
+      "step": 1602
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2679592051898685e-06,
+      "loss": 0.2312,
+      "step": 1604
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.244588730015264e-06,
+      "loss": 0.2428,
+      "step": 1606
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.221242109458043e-06,
+      "loss": 0.2551,
+      "step": 1608
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197919491817459e-06,
+      "loss": 0.2715,
+      "step": 1610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.174621025240307e-06,
+      "loss": 0.2359,
+      "step": 1612
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.151346857719964e-06,
+      "loss": 0.2671,
+      "step": 1614
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.128097137095458e-06,
+      "loss": 0.1967,
+      "step": 1616
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104872011050534e-06,
+      "loss": 0.2563,
+      "step": 1618
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.081671627112704e-06,
+      "loss": 0.2613,
+      "step": 1620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0584961326523285e-06,
+      "loss": 0.2962,
+      "step": 1622
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0353456748816545e-06,
+      "loss": 0.2087,
+      "step": 1624
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.012220400853899e-06,
+      "loss": 0.2902,
+      "step": 1626
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.989120457462314e-06,
+      "loss": 0.2565,
+      "step": 1628
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9660459914392465e-06,
+      "loss": 0.243,
+      "step": 1630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.942997149355208e-06,
+      "loss": 0.2895,
+      "step": 1632
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9199740776179494e-06,
+      "loss": 0.3008,
+      "step": 1634
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.89697692247152e-06,
+      "loss": 0.2124,
+      "step": 1636
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.874005829995358e-06,
+      "loss": 0.2094,
+      "step": 1638
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.851060946103334e-06,
+      "loss": 0.2355,
+      "step": 1640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.828142416542852e-06,
+      "loss": 0.304,
+      "step": 1642
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8052503868939005e-06,
+      "loss": 0.213,
+      "step": 1644
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.782385002568153e-06,
+      "loss": 0.2471,
+      "step": 1646
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.759546408808019e-06,
+      "loss": 0.2432,
+      "step": 1648
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.736734750685737e-06,
+      "loss": 0.2565,
+      "step": 1650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.713950173102441e-06,
+      "loss": 0.2502,
+      "step": 1652
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.691192820787266e-06,
+      "loss": 0.2105,
+      "step": 1654
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6684628382963905e-06,
+      "loss": 0.2437,
+      "step": 1656
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.645760370012149e-06,
+      "loss": 0.2149,
+      "step": 1658
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.623085560142099e-06,
+      "loss": 0.29,
+      "step": 1660
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.60043855271811e-06,
+      "loss": 0.2718,
+      "step": 1662
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.577819491595457e-06,
+      "loss": 0.2147,
+      "step": 1664
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.555228520451891e-06,
+      "loss": 0.1925,
+      "step": 1666
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.53266578278673e-06,
+      "loss": 0.2404,
+      "step": 1668
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.510131421919955e-06,
+      "loss": 0.2405,
+      "step": 1670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.487625580991303e-06,
+      "loss": 0.1999,
+      "step": 1672
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.465148402959339e-06,
+      "loss": 0.2185,
+      "step": 1674
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.442700030600565e-06,
+      "loss": 0.3006,
+      "step": 1676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.420280606508503e-06,
+      "loss": 0.2406,
+      "step": 1678
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.397890273092807e-06,
+      "loss": 0.2317,
+      "step": 1680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.375529172578329e-06,
+      "loss": 0.2616,
+      "step": 1682
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.353197447004239e-06,
+      "loss": 0.1408,
+      "step": 1684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.33089523822311e-06,
+      "loss": 0.1778,
+      "step": 1686
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.308622687900038e-06,
+      "loss": 0.1207,
+      "step": 1688
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.286379937511707e-06,
+      "loss": 0.1327,
+      "step": 1690
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.264167128345523e-06,
+      "loss": 0.125,
+      "step": 1692
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.241984401498693e-06,
+      "loss": 0.17,
+      "step": 1694
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.219831897877353e-06,
+      "loss": 0.159,
+      "step": 1696
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.197709758195648e-06,
+      "loss": 0.1156,
+      "step": 1698
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.175618122974851e-06,
+      "loss": 0.1356,
+      "step": 1700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.153557132542473e-06,
+      "loss": 0.1645,
+      "step": 1702
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.131526927031356e-06,
+      "loss": 0.1626,
+      "step": 1704
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.109527646378815e-06,
+      "loss": 0.1278,
+      "step": 1706
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.087559430325708e-06,
+      "loss": 0.136,
+      "step": 1708
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0656224184155764e-06,
+      "loss": 0.1141,
+      "step": 1710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.043716749993757e-06,
+      "loss": 0.1444,
+      "step": 1712
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.02184256420648e-06,
+      "loss": 0.1151,
+      "step": 1714
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.1474,
+      "step": 1716
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.978189196119716e-06,
+      "loss": 0.121,
+      "step": 1718
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9564102911092646e-06,
+      "loss": 0.1284,
+      "step": 1720
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.934663423309685e-06,
+      "loss": 0.153,
+      "step": 1722
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.912948730858492e-06,
+      "loss": 0.145,
+      "step": 1724
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.891266351688829e-06,
+      "loss": 0.1167,
+      "step": 1726
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.869616423528588e-06,
+      "loss": 0.1338,
+      "step": 1728
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.847999083899522e-06,
+      "loss": 0.1208,
+      "step": 1730
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.826414470116382e-06,
+      "loss": 0.1403,
+      "step": 1732
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.804862719286044e-06,
+      "loss": 0.1463,
+      "step": 1734
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.783343968306631e-06,
+      "loss": 0.1276,
+      "step": 1736
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.7618583538666605e-06,
+      "loss": 0.1242,
+      "step": 1738
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.740406012444153e-06,
+      "loss": 0.1402,
+      "step": 1740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.718987080305778e-06,
+      "loss": 0.3846,
+      "step": 1742
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.697601693505996e-06,
+      "loss": 0.161,
+      "step": 1744
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.6762499878861764e-06,
+      "loss": 0.1102,
+      "step": 1746
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.654932099073746e-06,
+      "loss": 0.1343,
+      "step": 1748
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.633648162481326e-06,
+      "loss": 0.146,
+      "step": 1750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.612398313305867e-06,
+      "loss": 0.1533,
+      "step": 1752
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.5911826865277975e-06,
+      "loss": 0.1346,
+      "step": 1754
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.570001416910168e-06,
+      "loss": 0.1579,
+      "step": 1756
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.548854638997778e-06,
+      "loss": 0.1382,
+      "step": 1758
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.527742487116349e-06,
+      "loss": 0.133,
+      "step": 1760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.506665095371642e-06,
+      "loss": 0.121,
+      "step": 1762
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.485622597648624e-06,
+      "loss": 0.1225,
+      "step": 1764
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.464615127610615e-06,
+      "loss": 0.1139,
+      "step": 1766
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.443642818698434e-06,
+      "loss": 0.1329,
+      "step": 1768
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.4227058041295515e-06,
+      "loss": 0.1131,
+      "step": 1770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.401804216897258e-06,
+      "loss": 0.1301,
+      "step": 1772
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.380938189769791e-06,
+      "loss": 0.1443,
+      "step": 1774
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3601078552895245e-06,
+      "loss": 0.1306,
+      "step": 1776
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.339313345772098e-06,
+      "loss": 0.1501,
+      "step": 1778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.318554793305592e-06,
+      "loss": 0.1076,
+      "step": 1780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.297832329749687e-06,
+      "loss": 0.1671,
+      "step": 1782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.277146086734823e-06,
+      "loss": 0.1364,
+      "step": 1784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2564961956613605e-06,
+      "loss": 0.1347,
+      "step": 1786
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.235882787698763e-06,
+      "loss": 0.1667,
+      "step": 1788
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2153059937847355e-06,
+      "loss": 0.1366,
+      "step": 1790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.194765944624423e-06,
+      "loss": 0.1142,
+      "step": 1792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.174262770689552e-06,
+      "loss": 0.1188,
+      "step": 1794
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.153796602217623e-06,
+      "loss": 0.1068,
+      "step": 1796
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.133367569211074e-06,
+      "loss": 0.1359,
+      "step": 1798
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.112975801436454e-06,
+      "loss": 0.1584,
+      "step": 1800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.092621428423601e-06,
+      "loss": 0.1308,
+      "step": 1802
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.07230457946482e-06,
+      "loss": 0.1324,
+      "step": 1804
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.052025383614061e-06,
+      "loss": 0.1339,
+      "step": 1806
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.031783969686105e-06,
+      "loss": 0.1178,
+      "step": 1808
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.011580466255729e-06,
+      "loss": 0.127,
+      "step": 1810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.991415001656906e-06,
+      "loss": 0.1334,
+      "step": 1812
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.971287703981982e-06,
+      "loss": 0.1201,
+      "step": 1814
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9511987010808635e-06,
+      "loss": 0.1435,
+      "step": 1816
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.931148120560211e-06,
+      "loss": 0.1394,
+      "step": 1818
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.911136089782613e-06,
+      "loss": 0.1229,
+      "step": 1820
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8911627358658e-06,
+      "loss": 0.123,
+      "step": 1822
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.871228185681822e-06,
+      "loss": 0.1433,
+      "step": 1824
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.8513325658562395e-06,
+      "loss": 0.1238,
+      "step": 1826
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.831476002767327e-06,
+      "loss": 0.1456,
+      "step": 1828
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.811658622545268e-06,
+      "loss": 0.1478,
+      "step": 1830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7918805510713553e-06,
+      "loss": 0.1269,
+      "step": 1832
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7721419139771886e-06,
+      "loss": 0.1379,
+      "step": 1834
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7524428366438757e-06,
+      "loss": 0.1365,
+      "step": 1836
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7327834442012433e-06,
+      "loss": 0.1263,
+      "step": 1838
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.7131638615270404e-06,
+      "loss": 0.1055,
+      "step": 1840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6935842132461307e-06,
+      "loss": 0.1053,
+      "step": 1842
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6740446237297177e-06,
+      "loss": 0.1259,
+      "step": 1844
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6545452170945496e-06,
+      "loss": 0.1428,
+      "step": 1846
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.635086117202128e-06,
+      "loss": 0.1375,
+      "step": 1848
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6156674476579266e-06,
+      "loss": 0.1469,
+      "step": 1850
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5962893318105963e-06,
+      "loss": 0.123,
+      "step": 1852
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.576951892751197e-06,
+      "loss": 0.128,
+      "step": 1854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5576552533124074e-06,
+      "loss": 0.1403,
+      "step": 1856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.538399536067736e-06,
+      "loss": 0.1392,
+      "step": 1858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5191848633307545e-06,
+      "loss": 0.1259,
+      "step": 1860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.5000113571543183e-06,
+      "loss": 0.1252,
+      "step": 1862
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.480879139329789e-06,
+      "loss": 0.1641,
+      "step": 1864
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4617883313862633e-06,
+      "loss": 0.1396,
+      "step": 1866
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.4427390545897955e-06,
+      "loss": 0.1284,
+      "step": 1868
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.423731429942636e-06,
+      "loss": 0.1255,
+      "step": 1870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.4047655781824605e-06,
+      "loss": 0.1381,
+      "step": 1872
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3858416197815947e-06,
+      "loss": 0.1587,
+      "step": 1874
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3669596749462562e-06,
+      "loss": 0.1148,
+      "step": 1876
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.3481198636157908e-06,
+      "loss": 0.1187,
+      "step": 1878
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3293223054619073e-06,
+      "loss": 0.1328,
+      "step": 1880
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.3105671198879243e-06,
+      "loss": 0.1166,
+      "step": 1882
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2918544260279985e-06,
+      "loss": 0.133,
+      "step": 1884
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2731843427463894e-06,
+      "loss": 0.127,
+      "step": 1886
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.254556988636678e-06,
+      "loss": 0.1678,
+      "step": 1888
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2359724820210394e-06,
+      "loss": 0.1156,
+      "step": 1890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.2174309409494675e-06,
+      "loss": 0.1384,
+      "step": 1892
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.198932483199041e-06,
+      "loss": 0.1324,
+      "step": 1894
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.180477226273172e-06,
+      "loss": 0.1498,
+      "step": 1896
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.162065287400855e-06,
+      "loss": 0.1482,
+      "step": 1898
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1436967835359245e-06,
+      "loss": 0.1001,
+      "step": 1900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1253718313563207e-06,
+      "loss": 0.1328,
+      "step": 1902
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.1070905472633307e-06,
+      "loss": 0.1343,
+      "step": 1904
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0888530473808677e-06,
+      "loss": 0.1721,
+      "step": 1906
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.070659447554719e-06,
+      "loss": 0.1211,
+      "step": 1908
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.052509863351818e-06,
+      "loss": 0.1267,
+      "step": 1910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.0344044100595073e-06,
+      "loss": 0.1257,
+      "step": 1912
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.016343202684807e-06,
+      "loss": 0.1769,
+      "step": 1914
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9983263559536813e-06,
+      "loss": 0.1398,
+      "step": 1916
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9803539843103226e-06,
+      "loss": 0.0818,
+      "step": 1918
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.962426201916402e-06,
+      "loss": 0.1552,
+      "step": 1920
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9445431226503683e-06,
+      "loss": 0.1296,
+      "step": 1922
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.926704860106706e-06,
+      "loss": 0.1082,
+      "step": 1924
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9089115275952217e-06,
+      "loss": 0.143,
+      "step": 1926
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.891163238140323e-06,
+      "loss": 0.1019,
+      "step": 1928
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8734601044803056e-06,
+      "loss": 0.1256,
+      "step": 1930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.855802239066623e-06,
+      "loss": 0.1159,
+      "step": 1932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8381897540631964e-06,
+      "loss": 0.1187,
+      "step": 1934
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.820622761345676e-06,
+      "loss": 0.1375,
+      "step": 1936
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.8031013725007415e-06,
+      "loss": 0.1305,
+      "step": 1938
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.785625698825406e-06,
+      "loss": 0.1443,
+      "step": 1940
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.768195851326285e-06,
+      "loss": 0.1351,
+      "step": 1942
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.750811940718906e-06,
+      "loss": 0.1378,
+      "step": 1944
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.733474077427004e-06,
+      "loss": 0.0981,
+      "step": 1946
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.716182371581814e-06,
+      "loss": 0.146,
+      "step": 1948
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6989369330213865e-06,
+      "loss": 0.1286,
+      "step": 1950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.681737871289869e-06,
+      "loss": 0.1551,
+      "step": 1952
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6645852956368214e-06,
+      "loss": 0.1166,
+      "step": 1954
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.647479315016528e-06,
+      "loss": 0.1181,
+      "step": 1956
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6304200380872913e-06,
+      "loss": 0.1341,
+      "step": 1958
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.61340757321075e-06,
+      "loss": 0.1196,
+      "step": 1960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.596442028451194e-06,
+      "loss": 0.1364,
+      "step": 1962
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.579523511574864e-06,
+      "loss": 0.1209,
+      "step": 1964
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.56265213004929e-06,
+      "loss": 0.1174,
+      "step": 1966
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5458279910425865e-06,
+      "loss": 0.1383,
+      "step": 1968
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5290512014227774e-06,
+      "loss": 0.1044,
+      "step": 1970
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.5123218677571313e-06,
+      "loss": 0.1163,
+      "step": 1972
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4956400963114647e-06,
+      "loss": 0.137,
+      "step": 1974
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.479005993049478e-06,
+      "loss": 0.1591,
+      "step": 1976
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4624196636320795e-06,
+      "loss": 0.137,
+      "step": 1978
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.445881213416713e-06,
+      "loss": 0.1583,
+      "step": 1980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.429390747456699e-06,
+      "loss": 0.1252,
+      "step": 1982
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.412948370500551e-06,
+      "loss": 0.1552,
+      "step": 1984
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3965541869913188e-06,
+      "loss": 0.1481,
+      "step": 1986
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3802083010659238e-06,
+      "loss": 0.1243,
+      "step": 1988
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3639108165545057e-06,
+      "loss": 0.1273,
+      "step": 1990
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3476618369797457e-06,
+      "loss": 0.1403,
+      "step": 1992
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.331461465556222e-06,
+      "loss": 0.1391,
+      "step": 1994
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.315309805189748e-06,
+      "loss": 0.1376,
+      "step": 1996
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.299206958476731e-06,
+      "loss": 0.1253,
+      "step": 1998
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2831530277034985e-06,
+      "loss": 0.131,
+      "step": 2000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2671481148456685e-06,
+      "loss": 0.1377,
+      "step": 2002
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.251192321567488e-06,
+      "loss": 0.1077,
+      "step": 2004
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.235285749221201e-06,
+      "loss": 0.1253,
+      "step": 2006
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.219428498846393e-06,
+      "loss": 0.1271,
+      "step": 2008
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2036206711693508e-06,
+      "loss": 0.1449,
+      "step": 2010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1878623666024233e-06,
+      "loss": 0.1024,
+      "step": 2012
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1721536852433976e-06,
+      "loss": 0.1141,
+      "step": 2014
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1564947268748382e-06,
+      "loss": 0.1023,
+      "step": 2016
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1408855909634696e-06,
+      "loss": 0.1113,
+      "step": 2018
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.125326376659539e-06,
+      "loss": 0.1467,
+      "step": 2020
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.1098171827961965e-06,
+      "loss": 0.1194,
+      "step": 2022
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.094358107888852e-06,
+      "loss": 0.1676,
+      "step": 2024
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0789492501345553e-06,
+      "loss": 0.1367,
+      "step": 2026
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0635907074113737e-06,
+      "loss": 0.1231,
+      "step": 2028
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0482825772777804e-06,
+      "loss": 0.1626,
+      "step": 2030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.0330249569720116e-06,
+      "loss": 0.1117,
+      "step": 2032
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.0178179434114674e-06,
+      "loss": 0.1303,
+      "step": 2034
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.00266163319209e-06,
+      "loss": 0.1336,
+      "step": 2036
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9875561225877482e-06,
+      "loss": 0.1143,
+      "step": 2038
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.972501507549637e-06,
+      "loss": 0.1449,
+      "step": 2040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.957497883705649e-06,
+      "loss": 0.1331,
+      "step": 2042
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9425453463597798e-06,
+      "loss": 0.0957,
+      "step": 2044
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.927643990491528e-06,
+      "loss": 0.1182,
+      "step": 2046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.912793910755275e-06,
+      "loss": 0.1394,
+      "step": 2048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8979952014796954e-06,
+      "loss": 0.1155,
+      "step": 2050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.883247956667157e-06,
+      "loss": 0.1681,
+      "step": 2052
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8685522699931169e-06,
+      "loss": 0.1517,
+      "step": 2054
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8539082348055427e-06,
+      "loss": 0.1491,
+      "step": 2056
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.839315944124298e-06,
+      "loss": 0.1276,
+      "step": 2058
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8247754906405624e-06,
+      "loss": 0.1343,
+      "step": 2060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8102869667162494e-06,
+      "loss": 0.1477,
+      "step": 2062
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7958504643834062e-06,
+      "loss": 0.1584,
+      "step": 2064
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7814660753436386e-06,
+      "loss": 0.1316,
+      "step": 2066
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7671338909675218e-06,
+      "loss": 0.1373,
+      "step": 2068
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7528540022940288e-06,
+      "loss": 0.131,
+      "step": 2070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7386265000299385e-06,
+      "loss": 0.1206,
+      "step": 2072
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7244514745492813e-06,
+      "loss": 0.117,
+      "step": 2074
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.71032901589274e-06,
+      "loss": 0.1368,
+      "step": 2076
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6962592137670897e-06,
+      "loss": 0.1176,
+      "step": 2078
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6822421575446378e-06,
+      "loss": 0.1501,
+      "step": 2080
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6682779362626378e-06,
+      "loss": 0.1326,
+      "step": 2082
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6543666386227343e-06,
+      "loss": 0.1357,
+      "step": 2084
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6405083529903954e-06,
+      "loss": 0.1039,
+      "step": 2086
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.6267031673943546e-06,
+      "loss": 0.1407,
+      "step": 2088
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6129511695260558e-06,
+      "loss": 0.1312,
+      "step": 2090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5992524467390858e-06,
+      "loss": 0.1198,
+      "step": 2092
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5856070860486205e-06,
+      "loss": 0.1091,
+      "step": 2094
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5720151741308875e-06,
+      "loss": 0.119,
+      "step": 2096
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5584767973225967e-06,
+      "loss": 0.1316,
+      "step": 2098
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.544992041620398e-06,
+      "loss": 0.1108,
+      "step": 2100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.531560992680341e-06,
+      "loss": 0.1267,
+      "step": 2102
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5181837358173223e-06,
+      "loss": 0.1292,
+      "step": 2104
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.5048603560045549e-06,
+      "loss": 0.124,
+      "step": 2106
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4915909378730143e-06,
+      "loss": 0.1466,
+      "step": 2108
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4783755657109079e-06,
+      "loss": 0.103,
+      "step": 2110
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4652143234631465e-06,
+      "loss": 0.1478,
+      "step": 2112
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.4521072947307957e-06,
+      "loss": 0.1196,
+      "step": 2114
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.4390545627705588e-06,
+      "loss": 0.1203,
+      "step": 2116
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.426056210494241e-06,
+      "loss": 0.125,
+      "step": 2118
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.413112320468223e-06,
+      "loss": 0.1612,
+      "step": 2120
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.400222974912936e-06,
+      "loss": 0.1226,
+      "step": 2122
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3873882557023488e-06,
+      "loss": 0.1304,
+      "step": 2124
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3746082443634311e-06,
+      "loss": 0.1172,
+      "step": 2126
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.361883022075653e-06,
+      "loss": 0.1441,
+      "step": 2128
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.3492126696704544e-06,
+      "loss": 0.1232,
+      "step": 2130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3365972676307403e-06,
+      "loss": 0.1127,
+      "step": 2132
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3240368960903671e-06,
+      "loss": 0.1298,
+      "step": 2134
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3115316348336348e-06,
+      "loss": 0.1358,
+      "step": 2136
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2990815632947763e-06,
+      "loss": 0.1689,
+      "step": 2138
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2866867605574628e-06,
+      "loss": 0.1101,
+      "step": 2140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2743473053542842e-06,
+      "loss": 0.1308,
+      "step": 2142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.262063276066272e-06,
+      "loss": 0.1472,
+      "step": 2144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.2498347507223763e-06,
+      "loss": 0.1298,
+      "step": 2146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.237661806998991e-06,
+      "loss": 0.1323,
+      "step": 2148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2255445222194462e-06,
+      "loss": 0.0947,
+      "step": 2150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2134829733535269e-06,
+      "loss": 0.1199,
+      "step": 2152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2014772370169747e-06,
+      "loss": 0.1284,
+      "step": 2154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1895273894710157e-06,
+      "loss": 0.1323,
+      "step": 2156
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.177633506621857e-06,
+      "loss": 0.1188,
+      "step": 2158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1657956640202217e-06,
+      "loss": 0.1448,
+      "step": 2160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.1540139368608572e-06,
+      "loss": 0.1819,
+      "step": 2162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.142288399982061e-06,
+      "loss": 0.2198,
+      "step": 2164
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1306191278652112e-06,
+      "loss": 0.1341,
+      "step": 2166
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1190061946342835e-06,
+      "loss": 0.2365,
+      "step": 2168
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1074496740553853e-06,
+      "loss": 0.1264,
+      "step": 2170
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0959496395362946e-06,
+      "loss": 0.1328,
+      "step": 2172
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0845061641259757e-06,
+      "loss": 0.1076,
+      "step": 2174
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0731193205141354e-06,
+      "loss": 0.1372,
+      "step": 2176
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.0617891810307458e-06,
+      "loss": 0.1536,
+      "step": 2178
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.050515817645591e-06,
+      "loss": 0.1243,
+      "step": 2180
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.039299301967811e-06,
+      "loss": 0.2138,
+      "step": 2182
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0281397052454457e-06,
+      "loss": 0.1357,
+      "step": 2184
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0170370983649792e-06,
+      "loss": 0.1623,
+      "step": 2186
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.005991551850899e-06,
+      "loss": 0.1314,
+      "step": 2188
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.950031358652313e-07,
+      "loss": 0.1163,
+      "step": 2190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.84071920207118e-07,
+      "loss": 0.1045,
+      "step": 2192
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.73197974312351e-07,
+      "loss": 0.1449,
+      "step": 2194
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.623813672529437e-07,
+      "loss": 0.1287,
+      "step": 2196
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.516221677366888e-07,
+      "loss": 0.1193,
+      "step": 2198
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.409204441067254e-07,
+      "loss": 0.1306,
+      "step": 2200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.302762643411e-07,
+      "loss": 0.1151,
+      "step": 2202
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.196896960523349e-07,
+      "loss": 0.1287,
+      "step": 2204
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.091608064870028e-07,
+      "loss": 0.099,
+      "step": 2206
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.986896625253006e-07,
+      "loss": 0.1151,
+      "step": 2208
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.882763306806163e-07,
+      "loss": 0.1466,
+      "step": 2210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.779208770991121e-07,
+      "loss": 0.1133,
+      "step": 2212
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.676233675593038e-07,
+      "loss": 0.157,
+      "step": 2214
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.573838674716461e-07,
+      "loss": 0.1184,
+      "step": 2216
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.472024418781099e-07,
+      "loss": 0.125,
+      "step": 2218
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.370791554517743e-07,
+      "loss": 0.146,
+      "step": 2220
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.270140724964159e-07,
+      "loss": 0.0981,
+      "step": 2222
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.170072569460996e-07,
+      "loss": 0.1288,
+      "step": 2224
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.070587723647705e-07,
+      "loss": 0.1714,
+      "step": 2226
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.971686819458502e-07,
+      "loss": 0.1147,
+      "step": 2228
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.873370485118381e-07,
+      "loss": 0.1307,
+      "step": 2230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.77563934513913e-07,
+      "loss": 0.1278,
+      "step": 2232
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.678494020315308e-07,
+      "loss": 0.118,
+      "step": 2234
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.581935127720352e-07,
+      "loss": 0.1289,
+      "step": 2236
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.485963280702646e-07,
+      "loss": 0.1139,
+      "step": 2238
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.390579088881655e-07,
+      "loss": 0.1164,
+      "step": 2240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.295783158143976e-07,
+      "loss": 0.0974,
+      "step": 2242
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.201576090639529e-07,
+      "loss": 0.1444,
+      "step": 2244
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.107958484777755e-07,
+      "loss": 0.1599,
+      "step": 2246
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.014930935223807e-07,
+      "loss": 0.1482,
+      "step": 2248
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.922494032894744e-07,
+      "loss": 0.1095,
+      "step": 2250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.830648364955772e-07,
+      "loss": 0.1398,
+      "step": 2252
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.739394514816622e-07,
+      "loss": 0.1333,
+      "step": 2254
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.648733062127643e-07,
+      "loss": 0.1209,
+      "step": 2256
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.558664582776341e-07,
+      "loss": 0.1019,
+      "step": 2258
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.469189648883567e-07,
+      "loss": 0.1099,
+      "step": 2260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.380308828799919e-07,
+      "loss": 0.1176,
+      "step": 2262
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.292022687102184e-07,
+      "loss": 0.1138,
+      "step": 2264
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.204331784589679e-07,
+      "loss": 0.1311,
+      "step": 2266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.117236678280736e-07,
+      "loss": 0.1296,
+      "step": 2268
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.1377,
+      "step": 2270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.9448360634207e-07,
+      "loss": 0.1579,
+      "step": 2272
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.859531649969563e-07,
+      "loss": 0.1133,
+      "step": 2274
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.774825222914948e-07,
+      "loss": 0.1324,
+      "step": 2276
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.690717320317595e-07,
+      "loss": 0.1227,
+      "step": 2278
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.60720847643641e-07,
+      "loss": 0.116,
+      "step": 2280
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.524299221724993e-07,
+      "loss": 0.1574,
+      "step": 2282
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.44199008282833e-07,
+      "loss": 0.1149,
+      "step": 2284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.360281582579474e-07,
+      "loss": 0.0964,
+      "step": 2286
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.279174239996132e-07,
+      "loss": 0.1096,
+      "step": 2288
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.198668570277443e-07,
+      "loss": 0.1395,
+      "step": 2290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.11876508480067e-07,
+      "loss": 0.1232,
+      "step": 2292
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.039464291117968e-07,
+      "loss": 0.1222,
+      "step": 2294
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.960766692953145e-07,
+      "loss": 0.16,
+      "step": 2296
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.882672790198473e-07,
+      "loss": 0.1558,
+      "step": 2298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.805183078911524e-07,
+      "loss": 0.1193,
+      "step": 2300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.728298051312008e-07,
+      "loss": 0.1342,
+      "step": 2302
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.652018195778629e-07,
+      "loss": 0.1598,
+      "step": 2304
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.576343996845989e-07,
+      "loss": 0.1324,
+      "step": 2306
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.5012759352015766e-07,
+      "loss": 0.0991,
+      "step": 2308
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.4268144876825846e-07,
+      "loss": 0.1399,
+      "step": 2310
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.352960127272987e-07,
+      "loss": 0.1098,
+      "step": 2312
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2797133231005207e-07,
+      "loss": 0.1343,
+      "step": 2314
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.207074540433631e-07,
+      "loss": 0.1038,
+      "step": 2316
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1350442406786317e-07,
+      "loss": 0.1445,
+      "step": 2318
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.063622881376683e-07,
+      "loss": 0.1484,
+      "step": 2320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9928109162008953e-07,
+      "loss": 0.1116,
+      "step": 2322
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.922608794953531e-07,
+      "loss": 0.1271,
+      "step": 2324
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8530169635630055e-07,
+      "loss": 0.1471,
+      "step": 2326
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.7840358640812036e-07,
+      "loss": 0.1074,
+      "step": 2328
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.715665934680546e-07,
+      "loss": 0.1571,
+      "step": 2330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.64790760965128e-07,
+      "loss": 0.0928,
+      "step": 2332
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.580761319398729e-07,
+      "loss": 0.1362,
+      "step": 2334
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.514227490440503e-07,
+      "loss": 0.1168,
+      "step": 2336
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.4483065454038123e-07,
+      "loss": 0.1497,
+      "step": 2338
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3829989030228163e-07,
+      "loss": 0.1244,
+      "step": 2340
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3183049781359e-07,
+      "loss": 0.1058,
+      "step": 2342
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.2542251816831237e-07,
+      "loss": 0.1158,
+      "step": 2344
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.190759920703512e-07,
+      "loss": 0.1208,
+      "step": 2346
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.127909598332535e-07,
+      "loss": 0.1214,
+      "step": 2348
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.065674613799574e-07,
+      "loss": 0.1258,
+      "step": 2350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.0040553624252844e-07,
+      "loss": 0.136,
+      "step": 2352
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.9430522356191814e-07,
+      "loss": 0.1553,
+      "step": 2354
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.88266562087709e-07,
+      "loss": 0.164,
+      "step": 2356
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.822895901778744e-07,
+      "loss": 0.1372,
+      "step": 2358
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.7637434579853016e-07,
+      "loss": 0.1196,
+      "step": 2360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.7052086652369356e-07,
+      "loss": 0.1351,
+      "step": 2362
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.6472918953504566e-07,
+      "loss": 0.1407,
+      "step": 2364
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.589993516216993e-07,
+      "loss": 0.1181,
+      "step": 2366
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.5333138917995714e-07,
+      "loss": 0.1484,
+      "step": 2368
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.477253382130862e-07,
+      "loss": 0.125,
+      "step": 2370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.4218123433108696e-07,
+      "loss": 0.1309,
+      "step": 2372
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.366991127504714e-07,
+      "loss": 0.1325,
+      "step": 2374
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.3127900829403305e-07,
+      "loss": 0.1297,
+      "step": 2376
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.259209553906272e-07,
+      "loss": 0.1242,
+      "step": 2378
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.2062498807495669e-07,
+      "loss": 0.155,
+      "step": 2380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1539113998735094e-07,
+      "loss": 0.1266,
+      "step": 2382
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.10219444373555e-07,
+      "loss": 0.1459,
+      "step": 2384
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.051099340845164e-07,
+      "loss": 0.1544,
+      "step": 2386
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.000626415761786e-07,
+      "loss": 0.1347,
+      "step": 2388
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9507759890927125e-07,
+      "loss": 0.146,
+      "step": 2390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.9015483774911249e-07,
+      "loss": 0.1342,
+      "step": 2392
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8529438936540022e-07,
+      "loss": 0.1166,
+      "step": 2394
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8049628463202128e-07,
+      "loss": 0.1313,
+      "step": 2396
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7576055402685034e-07,
+      "loss": 0.1251,
+      "step": 2398
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.710872276315556e-07,
+      "loss": 0.1077,
+      "step": 2400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.664763351314125e-07,
+      "loss": 0.1095,
+      "step": 2402
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.619279058151102e-07,
+      "loss": 0.0939,
+      "step": 2404
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5744196857456874e-07,
+      "loss": 0.1118,
+      "step": 2406
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5301855190475445e-07,
+      "loss": 0.1477,
+      "step": 2408
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4865768390349812e-07,
+      "loss": 0.1073,
+      "step": 2410
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4435939227131712e-07,
+      "loss": 0.1194,
+      "step": 2412
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4012370431124133e-07,
+      "loss": 0.1222,
+      "step": 2414
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3595064692863757e-07,
+      "loss": 0.1367,
+      "step": 2416
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3184024663103755e-07,
+      "loss": 0.1182,
+      "step": 2418
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.277925295279725e-07,
+      "loss": 0.1297,
+      "step": 2420
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2380752133080433e-07,
+      "loss": 0.1067,
+      "step": 2422
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.198852473525669e-07,
+      "loss": 0.1483,
+      "step": 2424
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1602573250779958e-07,
+      "loss": 0.1534,
+      "step": 2426
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1222900131239279e-07,
+      "loss": 0.1177,
+      "step": 2428
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0849507788343038e-07,
+      "loss": 0.2059,
+      "step": 2430
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0482398593903764e-07,
+      "loss": 0.1369,
+      "step": 2432
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0121574879823015e-07,
+      "loss": 0.1178,
+      "step": 2434
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.767038938076511e-08,
+      "loss": 0.1323,
+      "step": 2436
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.418793020699813e-08,
+      "loss": 0.1272,
+      "step": 2438
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.076839339773547e-08,
+      "loss": 0.112,
+      "step": 2440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.741180067409982e-08,
+      "loss": 0.1444,
+      "step": 2442
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.411817335738482e-08,
+      "loss": 0.1269,
+      "step": 2444
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.088753236892843e-08,
+      "loss": 0.1206,
+      "step": 2446
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.771989822997206e-08,
+      "loss": 0.1349,
+      "step": 2448
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.461529106153387e-08,
+      "loss": 0.1141,
+      "step": 2450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.15737305842823e-08,
+      "loss": 0.1265,
+      "step": 2452
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.859523611840612e-08,
+      "loss": 0.12,
+      "step": 2454
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.567982658349792e-08,
+      "loss": 0.1161,
+      "step": 2456
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.282752049842855e-08,
+      "loss": 0.1389,
+      "step": 2458
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.003833598123287e-08,
+      "loss": 0.1444,
+      "step": 2460
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.731229074899203e-08,
+      "loss": 0.11,
+      "step": 2462
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.464940211772574e-08,
+      "loss": 0.1272,
+      "step": 2464
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.204968700227242e-08,
+      "loss": 0.1016,
+      "step": 2466
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.951316191619593e-08,
+      "loss": 0.1521,
+      "step": 2468
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.703984297166564e-08,
+      "loss": 0.142,
+      "step": 2470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4629745879367634e-08,
+      "loss": 0.1034,
+      "step": 2472
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.228288594839369e-08,
+      "loss": 0.118,
+      "step": 2474
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.999927808615245e-08,
+      "loss": 0.1166,
+      "step": 2476
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.777893679827061e-08,
+      "loss": 0.1518,
+      "step": 2478
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.562187618849855e-08,
+      "loss": 0.1538,
+      "step": 2480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.352810995862932e-08,
+      "loss": 0.1053,
+      "step": 2482
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1497651408399774e-08,
+      "loss": 0.1095,
+      "step": 2484
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9530513435416243e-08,
+      "loss": 0.1574,
+      "step": 2486
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.762670853506677e-08,
+      "loss": 0.1147,
+      "step": 2488
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.578624880044567e-08,
+      "loss": 0.143,
+      "step": 2490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4009145922271327e-08,
+      "loss": 0.0848,
+      "step": 2492
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2295411188819616e-08,
+      "loss": 0.1291,
+      "step": 2494
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0645055485842837e-08,
+      "loss": 0.132,
+      "step": 2496
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9058089296509762e-08,
+      "loss": 0.106,
+      "step": 2498
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.753452270133238e-08,
+      "loss": 0.119,
+      "step": 2500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6074365378105915e-08,
+      "loss": 0.1213,
+      "step": 2502
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4677626601843353e-08,
+      "loss": 0.1392,
+      "step": 2504
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3344315244722128e-08,
+      "loss": 0.1121,
+      "step": 2506
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2074439776021962e-08,
+      "loss": 0.1163,
+      "step": 2508
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0868008262076013e-08,
+      "loss": 0.149,
+      "step": 2510
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.725028366214251e-09,
+      "loss": 0.1174,
+      "step": 2512
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.64550734872016e-09,
+      "loss": 0.1226,
+      "step": 2514
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.629452066783006e-09,
+      "loss": 0.1374,
+      "step": 2516
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.6768689744500796e-09,
+      "loss": 0.112,
+      "step": 2518
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.787764122592299e-09,
+      "loss": 0.1154,
+      "step": 2520
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.9621431588620096e-09,
+      "loss": 0.1275,
+      "step": 2522
+    },
+    {
+      "epoch": 3.0,
+      "step": 2523,
+      "total_flos": 2232048114991104.0,
+      "train_loss": 0.2896275484415448,
+      "train_runtime": 144327.3604,
+      "train_samples_per_second": 0.559,
+      "train_steps_per_second": 0.017
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 2523,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "total_flos": 2232048114991104.0,
+  "trial_name": null,
+  "trial_params": null
+}