diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,11776 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.9854247324071963,
+  "global_step": 980000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997458275043108e-05,
+      "loss": 6.084,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994916550086215e-05,
+      "loss": 4.6925,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.992374825129323e-05,
+      "loss": 4.2068,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989833100172431e-05,
+      "loss": 3.8478,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987291375215539e-05,
+      "loss": 3.6284,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984749650258646e-05,
+      "loss": 3.3725,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.982207925301754e-05,
+      "loss": 3.2057,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.979666200344861e-05,
+      "loss": 3.0135,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9771244753879695e-05,
+      "loss": 2.8507,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974582750431077e-05,
+      "loss": 2.7141,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9720410254741845e-05,
+      "loss": 2.7142,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969499300517292e-05,
+      "loss": 2.5984,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9669575755603995e-05,
+      "loss": 2.5191,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9644158506035074e-05,
+      "loss": 2.4221,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.961874125646615e-05,
+      "loss": 2.3656,
+      "step": 7500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.959332400689723e-05,
+      "loss": 2.3366,
+      "step": 8000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.95679067573283e-05,
+      "loss": 2.2823,
+      "step": 8500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.954248950775938e-05,
+      "loss": 2.2173,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951707225819045e-05,
+      "loss": 2.1778,
+      "step": 9500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949165500862154e-05,
+      "loss": 2.1375,
+      "step": 10000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.946623775905261e-05,
+      "loss": 2.1205,
+      "step": 10500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.944082050948369e-05,
+      "loss": 2.0824,
+      "step": 11000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.941540325991476e-05,
+      "loss": 2.065,
+      "step": 11500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.938998601034584e-05,
+      "loss": 2.0138,
+      "step": 12000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9364568760776917e-05,
+      "loss": 1.9739,
+      "step": 12500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9339151511207995e-05,
+      "loss": 1.9722,
+      "step": 13000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9313734261639073e-05,
+      "loss": 1.9094,
+      "step": 13500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9288317012070145e-05,
+      "loss": 1.9078,
+      "step": 14000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9262899762501224e-05,
+      "loss": 1.8806,
+      "step": 14500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9237482512932295e-05,
+      "loss": 1.8967,
+      "step": 15000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.921206526336338e-05,
+      "loss": 1.8345,
+      "step": 15500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.918664801379445e-05,
+      "loss": 1.8019,
+      "step": 16000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.916123076422553e-05,
+      "loss": 1.8188,
+      "step": 16500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.91358135146566e-05,
+      "loss": 1.7792,
+      "step": 17000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.911039626508768e-05,
+      "loss": 1.7658,
+      "step": 17500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.908497901551876e-05,
+      "loss": 1.7695,
+      "step": 18000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.905956176594984e-05,
+      "loss": 1.7261,
+      "step": 18500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.903414451638091e-05,
+      "loss": 1.7442,
+      "step": 19000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.900872726681199e-05,
+      "loss": 1.6924,
+      "step": 19500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8983310017243066e-05,
+      "loss": 1.6911,
+      "step": 20000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.895789276767414e-05,
+      "loss": 1.698,
+      "step": 20500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8932475518105217e-05,
+      "loss": 1.6869,
+      "step": 21000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8907058268536295e-05,
+      "loss": 1.658,
+      "step": 21500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8881641018967373e-05,
+      "loss": 1.6631,
+      "step": 22000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8856223769398445e-05,
+      "loss": 1.6475,
+      "step": 22500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8830806519829524e-05,
+      "loss": 1.6052,
+      "step": 23000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8805389270260595e-05,
+      "loss": 1.6436,
+      "step": 23500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877997202069168e-05,
+      "loss": 1.5989,
+      "step": 24000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875455477112275e-05,
+      "loss": 1.6462,
+      "step": 24500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872913752155383e-05,
+      "loss": 1.5783,
+      "step": 25000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87037202719849e-05,
+      "loss": 1.5768,
+      "step": 25500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.867830302241598e-05,
+      "loss": 1.6205,
+      "step": 26000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865288577284706e-05,
+      "loss": 1.5929,
+      "step": 26500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862746852327814e-05,
+      "loss": 1.5859,
+      "step": 27000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8602051273709216e-05,
+      "loss": 1.5663,
+      "step": 27500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857663402414029e-05,
+      "loss": 1.5626,
+      "step": 28000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8551216774571366e-05,
+      "loss": 1.5667,
+      "step": 28500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.852579952500244e-05,
+      "loss": 1.5311,
+      "step": 29000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.850038227543352e-05,
+      "loss": 1.5607,
+      "step": 29500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8474965025864595e-05,
+      "loss": 1.5504,
+      "step": 30000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8449547776295674e-05,
+      "loss": 1.5357,
+      "step": 30500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8424130526726745e-05,
+      "loss": 1.5756,
+      "step": 31000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8398713277157824e-05,
+      "loss": 1.5043,
+      "step": 31500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.83732960275889e-05,
+      "loss": 1.4885,
+      "step": 32000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.834787877801998e-05,
+      "loss": 1.5195,
+      "step": 32500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.832246152845106e-05,
+      "loss": 1.4871,
+      "step": 33000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.829704427888213e-05,
+      "loss": 1.4896,
+      "step": 33500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.827162702931321e-05,
+      "loss": 1.4791,
+      "step": 34000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.824620977974428e-05,
+      "loss": 1.4678,
+      "step": 34500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.822079253017536e-05,
+      "loss": 1.4817,
+      "step": 35000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.819537528060644e-05,
+      "loss": 1.4383,
+      "step": 35500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8169958031037516e-05,
+      "loss": 1.4782,
+      "step": 36000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.814454078146859e-05,
+      "loss": 1.4722,
+      "step": 36500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8119123531899666e-05,
+      "loss": 1.4466,
+      "step": 37000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8093706282330745e-05,
+      "loss": 1.4245,
+      "step": 37500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.806828903276182e-05,
+      "loss": 1.4572,
+      "step": 38000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8042871783192895e-05,
+      "loss": 1.4496,
+      "step": 38500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8017454533623974e-05,
+      "loss": 1.4244,
+      "step": 39000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.799203728405505e-05,
+      "loss": 1.4389,
+      "step": 39500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7966620034486124e-05,
+      "loss": 1.432,
+      "step": 40000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.79412027849172e-05,
+      "loss": 1.431,
+      "step": 40500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.791578553534828e-05,
+      "loss": 1.4335,
+      "step": 41000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.789036828577936e-05,
+      "loss": 1.4135,
+      "step": 41500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.786495103621043e-05,
+      "loss": 1.3965,
+      "step": 42000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.783953378664151e-05,
+      "loss": 1.4189,
+      "step": 42500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.781411653707258e-05,
+      "loss": 1.4053,
+      "step": 43000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7788699287503666e-05,
+      "loss": 1.4249,
+      "step": 43500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.776328203793474e-05,
+      "loss": 1.4328,
+      "step": 44000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7737864788365816e-05,
+      "loss": 1.3813,
+      "step": 44500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.771244753879689e-05,
+      "loss": 1.4115,
+      "step": 45000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7687030289227966e-05,
+      "loss": 1.3942,
+      "step": 45500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7661613039659045e-05,
+      "loss": 1.4045,
+      "step": 46000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.763619579009012e-05,
+      "loss": 1.3944,
+      "step": 46500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.76107785405212e-05,
+      "loss": 1.3701,
+      "step": 47000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.7585361290952274e-05,
+      "loss": 1.3892,
+      "step": 47500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.755994404138335e-05,
+      "loss": 1.3881,
+      "step": 48000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.7534526791814424e-05,
+      "loss": 1.4086,
+      "step": 48500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.750910954224551e-05,
+      "loss": 1.3873,
+      "step": 49000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.748369229267658e-05,
+      "loss": 1.38,
+      "step": 49500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.745827504310766e-05,
+      "loss": 1.3669,
+      "step": 50000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.743285779353873e-05,
+      "loss": 1.3219,
+      "step": 50500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.740744054396981e-05,
+      "loss": 1.3614,
+      "step": 51000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.738202329440089e-05,
+      "loss": 1.3752,
+      "step": 51500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7356606044831966e-05,
+      "loss": 1.3912,
+      "step": 52000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.7331188795263045e-05,
+      "loss": 1.3487,
+      "step": 52500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7305771545694116e-05,
+      "loss": 1.3566,
+      "step": 53000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7280354296125195e-05,
+      "loss": 1.3402,
+      "step": 53500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7254937046556266e-05,
+      "loss": 1.3424,
+      "step": 54000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7229519796987345e-05,
+      "loss": 1.3551,
+      "step": 54500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.720410254741842e-05,
+      "loss": 1.3385,
+      "step": 55000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.71786852978495e-05,
+      "loss": 1.3265,
+      "step": 55500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7153268048280574e-05,
+      "loss": 1.3236,
+      "step": 56000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.712785079871165e-05,
+      "loss": 1.3186,
+      "step": 56500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.710243354914273e-05,
+      "loss": 1.356,
+      "step": 57000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.707701629957381e-05,
+      "loss": 1.3236,
+      "step": 57500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.705159905000488e-05,
+      "loss": 1.3325,
+      "step": 58000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.702618180043596e-05,
+      "loss": 1.2949,
+      "step": 58500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.700076455086704e-05,
+      "loss": 1.3123,
+      "step": 59000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.697534730129811e-05,
+      "loss": 1.3211,
+      "step": 59500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.694993005172919e-05,
+      "loss": 1.3114,
+      "step": 60000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6924512802160266e-05,
+      "loss": 1.3321,
+      "step": 60500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6899095552591345e-05,
+      "loss": 1.3098,
+      "step": 61000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6873678303022416e-05,
+      "loss": 1.3289,
+      "step": 61500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6848261053453495e-05,
+      "loss": 1.2946,
+      "step": 62000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.6822843803884566e-05,
+      "loss": 1.3128,
+      "step": 62500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.679742655431565e-05,
+      "loss": 1.3231,
+      "step": 63000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.677200930474672e-05,
+      "loss": 1.323,
+      "step": 63500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.67465920551778e-05,
+      "loss": 1.3095,
+      "step": 64000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.6721174805608874e-05,
+      "loss": 1.3088,
+      "step": 64500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.669575755603995e-05,
+      "loss": 1.3054,
+      "step": 65000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.667034030647103e-05,
+      "loss": 1.3062,
+      "step": 65500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.664492305690211e-05,
+      "loss": 1.309,
+      "step": 66000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.661950580733319e-05,
+      "loss": 1.3151,
+      "step": 66500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.659408855776426e-05,
+      "loss": 1.2647,
+      "step": 67000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.656867130819534e-05,
+      "loss": 1.2767,
+      "step": 67500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.654325405862641e-05,
+      "loss": 1.2912,
+      "step": 68000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6517836809057494e-05,
+      "loss": 1.298,
+      "step": 68500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6492419559488566e-05,
+      "loss": 1.2743,
+      "step": 69000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6467002309919645e-05,
+      "loss": 1.2657,
+      "step": 69500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6441585060350716e-05,
+      "loss": 1.2979,
+      "step": 70000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.6416167810781795e-05,
+      "loss": 1.2808,
+      "step": 70500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.639075056121287e-05,
+      "loss": 1.2735,
+      "step": 71000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.636533331164395e-05,
+      "loss": 1.2782,
+      "step": 71500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.633991606207503e-05,
+      "loss": 1.2813,
+      "step": 72000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.63144988125061e-05,
+      "loss": 1.2316,
+      "step": 72500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.628908156293718e-05,
+      "loss": 1.2543,
+      "step": 73000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.626366431336825e-05,
+      "loss": 1.2771,
+      "step": 73500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.623824706379933e-05,
+      "loss": 1.2397,
+      "step": 74000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.621282981423041e-05,
+      "loss": 1.2692,
+      "step": 74500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.618741256466149e-05,
+      "loss": 1.2464,
+      "step": 75000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.616199531509256e-05,
+      "loss": 1.2739,
+      "step": 75500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.613657806552364e-05,
+      "loss": 1.2659,
+      "step": 76000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.611116081595471e-05,
+      "loss": 1.263,
+      "step": 76500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.6085743566385794e-05,
+      "loss": 1.253,
+      "step": 77000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6060326316816866e-05,
+      "loss": 1.2657,
+      "step": 77500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6034909067247945e-05,
+      "loss": 1.2782,
+      "step": 78000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.600949181767902e-05,
+      "loss": 1.2372,
+      "step": 78500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.5984074568110095e-05,
+      "loss": 1.22,
+      "step": 79000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.595865731854117e-05,
+      "loss": 1.2549,
+      "step": 79500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.593324006897225e-05,
+      "loss": 1.2684,
+      "step": 80000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.590782281940333e-05,
+      "loss": 1.2352,
+      "step": 80500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.58824055698344e-05,
+      "loss": 1.222,
+      "step": 81000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.585698832026548e-05,
+      "loss": 1.2205,
+      "step": 81500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.583157107069655e-05,
+      "loss": 1.2392,
+      "step": 82000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.580615382112764e-05,
+      "loss": 1.2246,
+      "step": 82500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.578073657155871e-05,
+      "loss": 1.2269,
+      "step": 83000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.575531932198979e-05,
+      "loss": 1.2329,
+      "step": 83500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.572990207242086e-05,
+      "loss": 1.21,
+      "step": 84000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.570448482285194e-05,
+      "loss": 1.2131,
+      "step": 84500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.5679067573283016e-05,
+      "loss": 1.2372,
+      "step": 85000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.5653650323714094e-05,
+      "loss": 1.2364,
+      "step": 85500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.562823307414517e-05,
+      "loss": 1.2311,
+      "step": 86000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.5602815824576245e-05,
+      "loss": 1.2202,
+      "step": 86500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.557739857500732e-05,
+      "loss": 1.2029,
+      "step": 87000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5551981325438395e-05,
+      "loss": 1.2067,
+      "step": 87500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.552656407586948e-05,
+      "loss": 1.2195,
+      "step": 88000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.550114682630055e-05,
+      "loss": 1.2262,
+      "step": 88500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.547572957673163e-05,
+      "loss": 1.2161,
+      "step": 89000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.54503123271627e-05,
+      "loss": 1.2139,
+      "step": 89500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.542489507759378e-05,
+      "loss": 1.2006,
+      "step": 90000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.539947782802486e-05,
+      "loss": 1.2308,
+      "step": 90500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.537406057845594e-05,
+      "loss": 1.1854,
+      "step": 91000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5348643328887016e-05,
+      "loss": 1.2261,
+      "step": 91500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.532322607931809e-05,
+      "loss": 1.1683,
+      "step": 92000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5297808829749166e-05,
+      "loss": 1.2081,
+      "step": 92500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.527239158018024e-05,
+      "loss": 1.1851,
+      "step": 93000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5246974330611316e-05,
+      "loss": 1.226,
+      "step": 93500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.5221557081042394e-05,
+      "loss": 1.2073,
+      "step": 94000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.519613983147347e-05,
+      "loss": 1.1998,
+      "step": 94500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5170722581904545e-05,
+      "loss": 1.2176,
+      "step": 95000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.514530533233562e-05,
+      "loss": 1.1956,
+      "step": 95500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.5119888082766695e-05,
+      "loss": 1.2167,
+      "step": 96000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.509447083319778e-05,
+      "loss": 1.2011,
+      "step": 96500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.506905358362885e-05,
+      "loss": 1.2115,
+      "step": 97000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.504363633405993e-05,
+      "loss": 1.1927,
+      "step": 97500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.501821908449101e-05,
+      "loss": 1.192,
+      "step": 98000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.499280183492208e-05,
+      "loss": 1.1802,
+      "step": 98500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.496738458535316e-05,
+      "loss": 1.1976,
+      "step": 99000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.494196733578424e-05,
+      "loss": 1.1976,
+      "step": 99500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4916550086215316e-05,
+      "loss": 1.2246,
+      "step": 100000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.489113283664639e-05,
+      "loss": 1.2115,
+      "step": 100500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4865715587077466e-05,
+      "loss": 1.1923,
+      "step": 101000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.484029833750854e-05,
+      "loss": 1.1731,
+      "step": 101500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.481488108793962e-05,
+      "loss": 1.1724,
+      "step": 102000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4789463838370694e-05,
+      "loss": 1.1902,
+      "step": 102500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.476404658880177e-05,
+      "loss": 1.1858,
+      "step": 103000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4738629339232845e-05,
+      "loss": 1.1789,
+      "step": 103500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.471321208966392e-05,
+      "loss": 1.1752,
+      "step": 104000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4687794840095e-05,
+      "loss": 1.2294,
+      "step": 104500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.466237759052608e-05,
+      "loss": 1.1889,
+      "step": 105000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.463696034095716e-05,
+      "loss": 1.1709,
+      "step": 105500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.461154309138823e-05,
+      "loss": 1.1834,
+      "step": 106000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.458612584181931e-05,
+      "loss": 1.1549,
+      "step": 106500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.456070859225038e-05,
+      "loss": 1.17,
+      "step": 107000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4535291342681466e-05,
+      "loss": 1.1852,
+      "step": 107500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.450987409311254e-05,
+      "loss": 1.1543,
+      "step": 108000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4484456843543616e-05,
+      "loss": 1.2015,
+      "step": 108500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.445903959397469e-05,
+      "loss": 1.1993,
+      "step": 109000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4433622344405766e-05,
+      "loss": 1.1285,
+      "step": 109500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4408205094836844e-05,
+      "loss": 1.1667,
+      "step": 110000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.438278784526792e-05,
+      "loss": 1.1504,
+      "step": 110500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.4357370595699e-05,
+      "loss": 1.1599,
+      "step": 111000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.433195334613007e-05,
+      "loss": 1.1756,
+      "step": 111500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.430653609656115e-05,
+      "loss": 1.154,
+      "step": 112000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.428111884699222e-05,
+      "loss": 1.1793,
+      "step": 112500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.42557015974233e-05,
+      "loss": 1.185,
+      "step": 113000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.423028434785438e-05,
+      "loss": 1.1764,
+      "step": 113500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.420486709828546e-05,
+      "loss": 1.1779,
+      "step": 114000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.417944984871653e-05,
+      "loss": 1.1517,
+      "step": 114500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.415403259914761e-05,
+      "loss": 1.1311,
+      "step": 115000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.412861534957868e-05,
+      "loss": 1.1504,
+      "step": 115500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.4103198100009766e-05,
+      "loss": 1.1475,
+      "step": 116000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.407778085044084e-05,
+      "loss": 1.1726,
+      "step": 116500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4052363600871916e-05,
+      "loss": 1.147,
+      "step": 117000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4026946351302994e-05,
+      "loss": 1.1571,
+      "step": 117500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4001529101734066e-05,
+      "loss": 1.1469,
+      "step": 118000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3976111852165144e-05,
+      "loss": 1.1374,
+      "step": 118500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.395069460259622e-05,
+      "loss": 1.1597,
+      "step": 119000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.39252773530273e-05,
+      "loss": 1.1406,
+      "step": 119500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.389986010345837e-05,
+      "loss": 1.119,
+      "step": 120000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.387444285388945e-05,
+      "loss": 1.1753,
+      "step": 120500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.384902560432052e-05,
+      "loss": 1.1358,
+      "step": 121000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.382360835475161e-05,
+      "loss": 1.1722,
+      "step": 121500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.379819110518268e-05,
+      "loss": 1.1193,
+      "step": 122000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.377277385561376e-05,
+      "loss": 1.1612,
+      "step": 122500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.374735660604483e-05,
+      "loss": 1.1679,
+      "step": 123000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.372193935647591e-05,
+      "loss": 1.1755,
+      "step": 123500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.369652210690699e-05,
+      "loss": 1.1227,
+      "step": 124000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3671104857338066e-05,
+      "loss": 1.1592,
+      "step": 124500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3645687607769144e-05,
+      "loss": 1.127,
+      "step": 125000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3620270358200216e-05,
+      "loss": 1.1644,
+      "step": 125500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3594853108631294e-05,
+      "loss": 1.1446,
+      "step": 126000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3569435859062366e-05,
+      "loss": 1.1356,
+      "step": 126500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.3544018609493444e-05,
+      "loss": 1.1154,
+      "step": 127000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.351860135992452e-05,
+      "loss": 1.1447,
+      "step": 127500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.34931841103556e-05,
+      "loss": 1.1277,
+      "step": 128000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.346776686078667e-05,
+      "loss": 1.1092,
+      "step": 128500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.344234961121775e-05,
+      "loss": 1.1491,
+      "step": 129000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.341693236164883e-05,
+      "loss": 1.1146,
+      "step": 129500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.339151511207991e-05,
+      "loss": 1.1263,
+      "step": 130000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.336609786251099e-05,
+      "loss": 1.1154,
+      "step": 130500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.334068061294206e-05,
+      "loss": 1.1464,
+      "step": 131000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.331526336337314e-05,
+      "loss": 1.1353,
+      "step": 131500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.328984611380421e-05,
+      "loss": 1.1447,
+      "step": 132000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.326442886423529e-05,
+      "loss": 1.1191,
+      "step": 132500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3239011614666366e-05,
+      "loss": 1.1525,
+      "step": 133000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3213594365097444e-05,
+      "loss": 1.1361,
+      "step": 133500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3188177115528516e-05,
+      "loss": 1.1292,
+      "step": 134000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3162759865959594e-05,
+      "loss": 1.1115,
+      "step": 134500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3137342616390666e-05,
+      "loss": 1.1384,
+      "step": 135000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.311192536682175e-05,
+      "loss": 1.1196,
+      "step": 135500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.308650811725282e-05,
+      "loss": 1.1335,
+      "step": 136000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30610908676839e-05,
+      "loss": 1.1276,
+      "step": 136500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.303567361811498e-05,
+      "loss": 1.1115,
+      "step": 137000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.301025636854605e-05,
+      "loss": 1.1086,
+      "step": 137500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.298483911897713e-05,
+      "loss": 1.1473,
+      "step": 138000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.295942186940821e-05,
+      "loss": 1.1089,
+      "step": 138500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.293400461983929e-05,
+      "loss": 1.1419,
+      "step": 139000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.290858737027036e-05,
+      "loss": 1.1291,
+      "step": 139500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.288317012070144e-05,
+      "loss": 1.1235,
+      "step": 140000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.285775287113251e-05,
+      "loss": 1.1109,
+      "step": 140500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2832335621563594e-05,
+      "loss": 1.1242,
+      "step": 141000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2806918371994666e-05,
+      "loss": 1.1142,
+      "step": 141500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2781501122425744e-05,
+      "loss": 1.108,
+      "step": 142000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2756083872856816e-05,
+      "loss": 1.1141,
+      "step": 142500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2730666623287894e-05,
+      "loss": 1.123,
+      "step": 143000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.270524937371897e-05,
+      "loss": 1.117,
+      "step": 143500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.267983212415005e-05,
+      "loss": 1.0937,
+      "step": 144000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.265441487458113e-05,
+      "loss": 1.1155,
+      "step": 144500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.26289976250122e-05,
+      "loss": 1.0759,
+      "step": 145000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.260358037544328e-05,
+      "loss": 1.113,
+      "step": 145500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.257816312587435e-05,
+      "loss": 1.0825,
+      "step": 146000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.255274587630543e-05,
+      "loss": 1.1138,
+      "step": 146500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.252732862673651e-05,
+      "loss": 1.1064,
+      "step": 147000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.250191137716759e-05,
+      "loss": 1.0978,
+      "step": 147500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.247649412759866e-05,
+      "loss": 1.1103,
+      "step": 148000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.245107687802974e-05,
+      "loss": 1.1052,
+      "step": 148500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2425659628460815e-05,
+      "loss": 1.1048,
+      "step": 149000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2400242378891894e-05,
+      "loss": 1.1024,
+      "step": 149500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.237482512932297e-05,
+      "loss": 1.1052,
+      "step": 150000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.2349407879754044e-05,
+      "loss": 1.0892,
+      "step": 150500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.232399063018512e-05,
+      "loss": 1.0934,
+      "step": 151000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.2298573380616194e-05,
+      "loss": 1.0674,
+      "step": 151500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.227315613104727e-05,
+      "loss": 1.0894,
+      "step": 152000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.224773888147835e-05,
+      "loss": 1.0652,
+      "step": 152500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.222232163190943e-05,
+      "loss": 1.0869,
+      "step": 153000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.21969043823405e-05,
+      "loss": 1.0769,
+      "step": 153500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.217148713277158e-05,
+      "loss": 1.08,
+      "step": 154000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.214606988320265e-05,
+      "loss": 1.0651,
+      "step": 154500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.212065263363374e-05,
+      "loss": 1.0691,
+      "step": 155000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.209523538406481e-05,
+      "loss": 1.0981,
+      "step": 155500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.206981813449589e-05,
+      "loss": 1.0722,
+      "step": 156000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.2044400884926965e-05,
+      "loss": 1.0944,
+      "step": 156500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.201898363535804e-05,
+      "loss": 1.0613,
+      "step": 157000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.1993566385789115e-05,
+      "loss": 1.0971,
+      "step": 157500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.1968149136220194e-05,
+      "loss": 1.0781,
+      "step": 158000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.194273188665127e-05,
+      "loss": 1.1003,
+      "step": 158500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.1917314637082344e-05,
+      "loss": 1.1027,
+      "step": 159000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.189189738751342e-05,
+      "loss": 1.0872,
+      "step": 159500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.1866480137944494e-05,
+      "loss": 1.0872,
+      "step": 160000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.184106288837558e-05,
+      "loss": 1.0766,
+      "step": 160500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.181564563880665e-05,
+      "loss": 1.0905,
+      "step": 161000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.179022838923773e-05,
+      "loss": 1.0786,
+      "step": 161500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.17648111396688e-05,
+      "loss": 1.051,
+      "step": 162000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.173939389009988e-05,
+      "loss": 1.0724,
+      "step": 162500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.171397664053096e-05,
+      "loss": 1.0752,
+      "step": 163000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.168855939096204e-05,
+      "loss": 1.0915,
+      "step": 163500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.1663142141393115e-05,
+      "loss": 1.0933,
+      "step": 164000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.163772489182419e-05,
+      "loss": 1.0878,
+      "step": 164500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.1612307642255265e-05,
+      "loss": 1.0679,
+      "step": 165000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.158689039268634e-05,
+      "loss": 1.0614,
+      "step": 165500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1561473143117415e-05,
+      "loss": 1.0977,
+      "step": 166000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1536055893548494e-05,
+      "loss": 1.0586,
+      "step": 166500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.151063864397957e-05,
+      "loss": 1.0809,
+      "step": 167000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.1485221394410644e-05,
+      "loss": 1.0594,
+      "step": 167500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.145980414484172e-05,
+      "loss": 1.0743,
+      "step": 168000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.1434386895272794e-05,
+      "loss": 1.0745,
+      "step": 168500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.140896964570388e-05,
+      "loss": 1.0587,
+      "step": 169000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.138355239613496e-05,
+      "loss": 1.0765,
+      "step": 169500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.135813514656603e-05,
+      "loss": 1.0381,
+      "step": 170000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.133271789699711e-05,
+      "loss": 1.0834,
+      "step": 170500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.130730064742818e-05,
+      "loss": 1.0873,
+      "step": 171000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.128188339785926e-05,
+      "loss": 1.0896,
+      "step": 171500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.125646614829034e-05,
+      "loss": 1.0398,
+      "step": 172000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1231048898721415e-05,
+      "loss": 1.0437,
+      "step": 172500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.120563164915249e-05,
+      "loss": 1.0858,
+      "step": 173000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1180214399583565e-05,
+      "loss": 1.0593,
+      "step": 173500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115479715001464e-05,
+      "loss": 1.0628,
+      "step": 174000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112937990044572e-05,
+      "loss": 1.0782,
+      "step": 174500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1103962650876794e-05,
+      "loss": 1.0786,
+      "step": 175000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107854540130787e-05,
+      "loss": 1.0837,
+      "step": 175500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.105312815173895e-05,
+      "loss": 1.049,
+      "step": 176000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.102771090217002e-05,
+      "loss": 1.0534,
+      "step": 176500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.10022936526011e-05,
+      "loss": 1.0772,
+      "step": 177000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.097687640303218e-05,
+      "loss": 1.0564,
+      "step": 177500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.095145915346326e-05,
+      "loss": 1.068,
+      "step": 178000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.092604190389433e-05,
+      "loss": 1.0445,
+      "step": 178500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.090062465432541e-05,
+      "loss": 1.083,
+      "step": 179000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.087520740475648e-05,
+      "loss": 1.0687,
+      "step": 179500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0849790155187565e-05,
+      "loss": 1.0618,
+      "step": 180000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.082437290561864e-05,
+      "loss": 1.0677,
+      "step": 180500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.0798955656049715e-05,
+      "loss": 1.0278,
+      "step": 181000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.077353840648079e-05,
+      "loss": 1.0622,
+      "step": 181500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.0748121156911865e-05,
+      "loss": 1.0243,
+      "step": 182000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.0722703907342944e-05,
+      "loss": 1.0516,
+      "step": 182500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.069728665777402e-05,
+      "loss": 1.0739,
+      "step": 183000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.06718694082051e-05,
+      "loss": 1.0586,
+      "step": 183500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.064645215863617e-05,
+      "loss": 1.0517,
+      "step": 184000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.062103490906725e-05,
+      "loss": 1.0687,
+      "step": 184500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.059561765949832e-05,
+      "loss": 1.0386,
+      "step": 185000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.05702004099294e-05,
+      "loss": 1.0836,
+      "step": 185500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.054478316036048e-05,
+      "loss": 1.0686,
+      "step": 186000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.051936591079156e-05,
+      "loss": 1.044,
+      "step": 186500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.049394866122263e-05,
+      "loss": 1.0519,
+      "step": 187000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.046853141165371e-05,
+      "loss": 1.0606,
+      "step": 187500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.044311416208478e-05,
+      "loss": 1.0325,
+      "step": 188000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0417696912515865e-05,
+      "loss": 1.0541,
+      "step": 188500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0392279662946943e-05,
+      "loss": 1.0592,
+      "step": 189000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0366862413378015e-05,
+      "loss": 1.0479,
+      "step": 189500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0341445163809094e-05,
+      "loss": 1.0156,
+      "step": 190000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.0316027914240165e-05,
+      "loss": 1.0185,
+      "step": 190500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.0290610664671244e-05,
+      "loss": 1.0439,
+      "step": 191000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.026519341510232e-05,
+      "loss": 1.0654,
+      "step": 191500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.02397761655334e-05,
+      "loss": 1.0403,
+      "step": 192000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.021435891596447e-05,
+      "loss": 1.0457,
+      "step": 192500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.018894166639555e-05,
+      "loss": 1.0766,
+      "step": 193000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.016352441682662e-05,
+      "loss": 1.0418,
+      "step": 193500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.013810716725771e-05,
+      "loss": 1.054,
+      "step": 194000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.011268991768878e-05,
+      "loss": 1.0215,
+      "step": 194500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.008727266811986e-05,
+      "loss": 1.038,
+      "step": 195000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0061855418550936e-05,
+      "loss": 1.0522,
+      "step": 195500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003643816898201e-05,
+      "loss": 1.0287,
+      "step": 196000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0011020919413087e-05,
+      "loss": 1.0367,
+      "step": 196500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9985603669844165e-05,
+      "loss": 1.0382,
+      "step": 197000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9960186420275244e-05,
+      "loss": 1.0342,
+      "step": 197500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9934769170706315e-05,
+      "loss": 1.0587,
+      "step": 198000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9909351921137394e-05,
+      "loss": 1.0363,
+      "step": 198500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9883934671568465e-05,
+      "loss": 1.0419,
+      "step": 199000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9858517421999544e-05,
+      "loss": 1.0222,
+      "step": 199500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.983310017243062e-05,
+      "loss": 1.0381,
+      "step": 200000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.98076829228617e-05,
+      "loss": 1.0432,
+      "step": 200500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.978226567329277e-05,
+      "loss": 1.0554,
+      "step": 201000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.975684842372385e-05,
+      "loss": 1.0281,
+      "step": 201500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.973143117415493e-05,
+      "loss": 1.0761,
+      "step": 202000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.970601392458601e-05,
+      "loss": 1.0476,
+      "step": 202500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9680596675017086e-05,
+      "loss": 1.0515,
+      "step": 203000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.965517942544816e-05,
+      "loss": 1.0426,
+      "step": 203500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9629762175879236e-05,
+      "loss": 1.0254,
+      "step": 204000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.960434492631031e-05,
+      "loss": 1.0364,
+      "step": 204500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9578927676741387e-05,
+      "loss": 1.0633,
+      "step": 205000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9553510427172465e-05,
+      "loss": 1.0115,
+      "step": 205500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9528093177603544e-05,
+      "loss": 1.0165,
+      "step": 206000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9502675928034615e-05,
+      "loss": 1.0573,
+      "step": 206500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9477258678465694e-05,
+      "loss": 1.011,
+      "step": 207000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9451841428896765e-05,
+      "loss": 1.0333,
+      "step": 207500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.942642417932785e-05,
+      "loss": 1.0166,
+      "step": 208000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.940100692975893e-05,
+      "loss": 1.0184,
+      "step": 208500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.937558968019e-05,
+      "loss": 1.048,
+      "step": 209000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.935017243062108e-05,
+      "loss": 1.0397,
+      "step": 209500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.932475518105215e-05,
+      "loss": 1.0178,
+      "step": 210000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.929933793148323e-05,
+      "loss": 1.0127,
+      "step": 210500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.927392068191431e-05,
+      "loss": 1.0211,
+      "step": 211000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9248503432345386e-05,
+      "loss": 1.0584,
+      "step": 211500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.922308618277646e-05,
+      "loss": 1.0269,
+      "step": 212000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9197668933207536e-05,
+      "loss": 1.0076,
+      "step": 212500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.917225168363861e-05,
+      "loss": 1.0161,
+      "step": 213000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.914683443406969e-05,
+      "loss": 1.0352,
+      "step": 213500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9121417184500765e-05,
+      "loss": 1.0237,
+      "step": 214000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9095999934931844e-05,
+      "loss": 1.0545,
+      "step": 214500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.907058268536292e-05,
+      "loss": 1.0175,
+      "step": 215000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9045165435793994e-05,
+      "loss": 1.0201,
+      "step": 215500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.901974818622507e-05,
+      "loss": 1.0244,
+      "step": 216000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.899433093665615e-05,
+      "loss": 1.0358,
+      "step": 216500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.896891368708723e-05,
+      "loss": 1.0505,
+      "step": 217000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89434964375183e-05,
+      "loss": 1.0371,
+      "step": 217500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.891807918794938e-05,
+      "loss": 1.0394,
+      "step": 218000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.889266193838045e-05,
+      "loss": 1.0114,
+      "step": 218500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.886724468881153e-05,
+      "loss": 1.0188,
+      "step": 219000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.884182743924261e-05,
+      "loss": 1.031,
+      "step": 219500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8816410189673686e-05,
+      "loss": 1.0585,
+      "step": 220000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.879099294010476e-05,
+      "loss": 1.0261,
+      "step": 220500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8765575690535836e-05,
+      "loss": 1.0124,
+      "step": 221000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.8740158440966915e-05,
+      "loss": 1.026,
+      "step": 221500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.871474119139799e-05,
+      "loss": 1.0264,
+      "step": 222000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.868932394182907e-05,
+      "loss": 1.002,
+      "step": 222500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8663906692260144e-05,
+      "loss": 1.022,
+      "step": 223000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.863848944269122e-05,
+      "loss": 0.9912,
+      "step": 223500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8613072193122294e-05,
+      "loss": 1.0268,
+      "step": 224000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858765494355337e-05,
+      "loss": 1.0219,
+      "step": 224500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.856223769398445e-05,
+      "loss": 1.0178,
+      "step": 225000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.853682044441553e-05,
+      "loss": 1.0228,
+      "step": 225500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.85114031948466e-05,
+      "loss": 0.9964,
+      "step": 226000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.848598594527768e-05,
+      "loss": 1.013,
+      "step": 226500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.846056869570875e-05,
+      "loss": 0.9925,
+      "step": 227000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8435151446139836e-05,
+      "loss": 1.0315,
+      "step": 227500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8409734196570915e-05,
+      "loss": 1.0126,
+      "step": 228000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8384316947001986e-05,
+      "loss": 1.0205,
+      "step": 228500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8358899697433065e-05,
+      "loss": 1.0329,
+      "step": 229000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.8333482447864136e-05,
+      "loss": 1.017,
+      "step": 229500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8308065198295215e-05,
+      "loss": 1.0306,
+      "step": 230000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.828264794872629e-05,
+      "loss": 1.0083,
+      "step": 230500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.825723069915737e-05,
+      "loss": 1.017,
+      "step": 231000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8231813449588444e-05,
+      "loss": 1.0099,
+      "step": 231500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.820639620001952e-05,
+      "loss": 1.0077,
+      "step": 232000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8180978950450594e-05,
+      "loss": 1.015,
+      "step": 232500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.815556170088168e-05,
+      "loss": 1.0036,
+      "step": 233000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.813014445131275e-05,
+      "loss": 1.015,
+      "step": 233500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.810472720174383e-05,
+      "loss": 1.0267,
+      "step": 234000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.807930995217491e-05,
+      "loss": 1.0024,
+      "step": 234500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.805389270260598e-05,
+      "loss": 1.0123,
+      "step": 235000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.802847545303706e-05,
+      "loss": 1.003,
+      "step": 235500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8003058203468136e-05,
+      "loss": 1.0308,
+      "step": 236000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7977640953899215e-05,
+      "loss": 0.9973,
+      "step": 236500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7952223704330286e-05,
+      "loss": 1.0182,
+      "step": 237000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7926806454761365e-05,
+      "loss": 1.0043,
+      "step": 237500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7901389205192436e-05,
+      "loss": 1.018,
+      "step": 238000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7875971955623515e-05,
+      "loss": 0.9994,
+      "step": 238500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.785055470605459e-05,
+      "loss": 1.0178,
+      "step": 239000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.782513745648567e-05,
+      "loss": 1.0293,
+      "step": 239500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7799720206916744e-05,
+      "loss": 1.0135,
+      "step": 240000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.777430295734782e-05,
+      "loss": 0.9965,
+      "step": 240500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.77488857077789e-05,
+      "loss": 1.0073,
+      "step": 241000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.772346845820998e-05,
+      "loss": 0.9967,
+      "step": 241500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.769805120864106e-05,
+      "loss": 0.9866,
+      "step": 242000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.767263395907213e-05,
+      "loss": 0.9946,
+      "step": 242500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.764721670950321e-05,
+      "loss": 1.022,
+      "step": 243000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762179945993428e-05,
+      "loss": 0.9928,
+      "step": 243500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.759638221036536e-05,
+      "loss": 1.0135,
+      "step": 244000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7570964960796436e-05,
+      "loss": 1.0125,
+      "step": 244500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7545547711227515e-05,
+      "loss": 1.0134,
+      "step": 245000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7520130461658586e-05,
+      "loss": 1.0075,
+      "step": 245500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7494713212089665e-05,
+      "loss": 0.9923,
+      "step": 246000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7469295962520736e-05,
+      "loss": 0.9816,
+      "step": 246500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.744387871295182e-05,
+      "loss": 0.9939,
+      "step": 247000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.74184614633829e-05,
+      "loss": 1.0022,
+      "step": 247500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.739304421381397e-05,
+      "loss": 0.9931,
+      "step": 248000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.736762696424505e-05,
+      "loss": 1.0011,
+      "step": 248500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.734220971467612e-05,
+      "loss": 1.011,
+      "step": 249000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.73167924651072e-05,
+      "loss": 0.9717,
+      "step": 249500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.729137521553828e-05,
+      "loss": 0.985,
+      "step": 250000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.726595796596936e-05,
+      "loss": 1.0036,
+      "step": 250500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.724054071640043e-05,
+      "loss": 1.0059,
+      "step": 251000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.721512346683151e-05,
+      "loss": 0.9749,
+      "step": 251500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.718970621726258e-05,
+      "loss": 0.9962,
+      "step": 252000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7164288967693664e-05,
+      "loss": 1.0007,
+      "step": 252500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7138871718124736e-05,
+      "loss": 0.9933,
+      "step": 253000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7113454468555815e-05,
+      "loss": 1.011,
+      "step": 253500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.708803721898689e-05,
+      "loss": 1.0118,
+      "step": 254000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.7062619969417965e-05,
+      "loss": 0.9985,
+      "step": 254500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.703720271984904e-05,
+      "loss": 0.9851,
+      "step": 255000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.701178547028012e-05,
+      "loss": 1.0026,
+      "step": 255500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.69863682207112e-05,
+      "loss": 0.9876,
+      "step": 256000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.696095097114227e-05,
+      "loss": 0.9853,
+      "step": 256500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.693553372157335e-05,
+      "loss": 0.9917,
+      "step": 257000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.691011647200442e-05,
+      "loss": 1.0115,
+      "step": 257500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.68846992224355e-05,
+      "loss": 1.0027,
+      "step": 258000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.685928197286658e-05,
+      "loss": 0.9872,
+      "step": 258500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.683386472329766e-05,
+      "loss": 0.9891,
+      "step": 259000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.680844747372873e-05,
+      "loss": 1.0136,
+      "step": 259500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.678303022415981e-05,
+      "loss": 1.0092,
+      "step": 260000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6757612974590886e-05,
+      "loss": 0.9798,
+      "step": 260500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.6732195725021964e-05,
+      "loss": 0.9841,
+      "step": 261000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.670677847545304e-05,
+      "loss": 1.004,
+      "step": 261500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.6681361225884115e-05,
+      "loss": 0.9789,
+      "step": 262000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.665594397631519e-05,
+      "loss": 1.0164,
+      "step": 262500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.6630526726746265e-05,
+      "loss": 0.9854,
+      "step": 263000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.660510947717734e-05,
+      "loss": 1.0258,
+      "step": 263500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.657969222760842e-05,
+      "loss": 0.99,
+      "step": 264000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.65542749780395e-05,
+      "loss": 0.9736,
+      "step": 264500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.652885772847057e-05,
+      "loss": 0.9871,
+      "step": 265000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.650344047890165e-05,
+      "loss": 0.9733,
+      "step": 265500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.647802322933272e-05,
+      "loss": 0.9894,
+      "step": 266000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.645260597976381e-05,
+      "loss": 0.9671,
+      "step": 266500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6427188730194886e-05,
+      "loss": 1.0113,
+      "step": 267000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.640177148062596e-05,
+      "loss": 0.9942,
+      "step": 267500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6376354231057036e-05,
+      "loss": 0.9765,
+      "step": 268000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.635093698148811e-05,
+      "loss": 0.9805,
+      "step": 268500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.6325519731919186e-05,
+      "loss": 0.9789,
+      "step": 269000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6300102482350264e-05,
+      "loss": 0.962,
+      "step": 269500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.627468523278134e-05,
+      "loss": 0.987,
+      "step": 270000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6249267983212415e-05,
+      "loss": 0.994,
+      "step": 270500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.622385073364349e-05,
+      "loss": 0.9791,
+      "step": 271000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.6198433484074565e-05,
+      "loss": 0.9802,
+      "step": 271500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.617301623450565e-05,
+      "loss": 0.9884,
+      "step": 272000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.614759898493672e-05,
+      "loss": 0.9666,
+      "step": 272500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.61221817353678e-05,
+      "loss": 0.9796,
+      "step": 273000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.609676448579888e-05,
+      "loss": 1.0113,
+      "step": 273500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.607134723622995e-05,
+      "loss": 0.9584,
+      "step": 274000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.604592998666103e-05,
+      "loss": 0.9843,
+      "step": 274500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.602051273709211e-05,
+      "loss": 0.9706,
+      "step": 275000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.5995095487523186e-05,
+      "loss": 0.9738,
+      "step": 275500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.596967823795426e-05,
+      "loss": 0.9695,
+      "step": 276000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.5944260988385336e-05,
+      "loss": 0.9862,
+      "step": 276500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.591884373881641e-05,
+      "loss": 0.9529,
+      "step": 277000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.5893426489247486e-05,
+      "loss": 0.9981,
+      "step": 277500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.5868009239678564e-05,
+      "loss": 0.982,
+      "step": 278000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.584259199010964e-05,
+      "loss": 0.9948,
+      "step": 278500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.5817174740540715e-05,
+      "loss": 0.9455,
+      "step": 279000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.579175749097179e-05,
+      "loss": 0.978,
+      "step": 279500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.576634024140287e-05,
+      "loss": 0.9803,
+      "step": 280000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.574092299183395e-05,
+      "loss": 0.9922,
+      "step": 280500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.571550574226503e-05,
+      "loss": 0.981,
+      "step": 281000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.56900884926961e-05,
+      "loss": 0.9839,
+      "step": 281500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.566467124312718e-05,
+      "loss": 0.9637,
+      "step": 282000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.563925399355825e-05,
+      "loss": 0.9769,
+      "step": 282500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.561383674398933e-05,
+      "loss": 0.9737,
+      "step": 283000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.558841949442041e-05,
+      "loss": 0.9913,
+      "step": 283500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.5563002244851486e-05,
+      "loss": 0.9876,
+      "step": 284000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.553758499528256e-05,
+      "loss": 0.9676,
+      "step": 284500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.5512167745713636e-05,
+      "loss": 0.9541,
+      "step": 285000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.548675049614471e-05,
+      "loss": 0.9777,
+      "step": 285500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.546133324657579e-05,
+      "loss": 0.9754,
+      "step": 286000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.543591599700687e-05,
+      "loss": 0.9899,
+      "step": 286500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.541049874743794e-05,
+      "loss": 0.9628,
+      "step": 287000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.538508149786902e-05,
+      "loss": 0.9673,
+      "step": 287500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.535966424830009e-05,
+      "loss": 0.9899,
+      "step": 288000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.533424699873117e-05,
+      "loss": 0.9645,
+      "step": 288500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.530882974916225e-05,
+      "loss": 0.961,
+      "step": 289000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.528341249959333e-05,
+      "loss": 0.9752,
+      "step": 289500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.52579952500244e-05,
+      "loss": 0.9851,
+      "step": 290000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.523257800045548e-05,
+      "loss": 0.9656,
+      "step": 290500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.520716075088655e-05,
+      "loss": 0.9924,
+      "step": 291000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.518174350131763e-05,
+      "loss": 0.9897,
+      "step": 291500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.515632625174871e-05,
+      "loss": 0.9757,
+      "step": 292000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5130909002179786e-05,
+      "loss": 0.9729,
+      "step": 292500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5105491752610864e-05,
+      "loss": 0.96,
+      "step": 293000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.5080074503041936e-05,
+      "loss": 0.9751,
+      "step": 293500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.5054657253473014e-05,
+      "loss": 0.9665,
+      "step": 294000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.502924000390409e-05,
+      "loss": 0.9662,
+      "step": 294500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.500382275433517e-05,
+      "loss": 0.9685,
+      "step": 295000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.497840550476624e-05,
+      "loss": 0.973,
+      "step": 295500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.495298825519732e-05,
+      "loss": 0.9773,
+      "step": 296000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.492757100562839e-05,
+      "loss": 0.9946,
+      "step": 296500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.490215375605947e-05,
+      "loss": 0.9578,
+      "step": 297000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.487673650649055e-05,
+      "loss": 0.9814,
+      "step": 297500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.485131925692163e-05,
+      "loss": 0.9831,
+      "step": 298000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.48259020073527e-05,
+      "loss": 0.9709,
+      "step": 298500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.480048475778378e-05,
+      "loss": 0.9673,
+      "step": 299000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.477506750821486e-05,
+      "loss": 0.9738,
+      "step": 299500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.4749650258645936e-05,
+      "loss": 0.9825,
+      "step": 300000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.4724233009077014e-05,
+      "loss": 0.978,
+      "step": 300500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.4698815759508086e-05,
+      "loss": 0.9607,
+      "step": 301000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.4673398509939164e-05,
+      "loss": 0.9859,
+      "step": 301500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.4647981260370236e-05,
+      "loss": 0.9984,
+      "step": 302000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.4622564010801314e-05,
+      "loss": 0.9776,
+      "step": 302500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.459714676123239e-05,
+      "loss": 0.9546,
+      "step": 303000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.457172951166347e-05,
+      "loss": 0.9661,
+      "step": 303500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.454631226209454e-05,
+      "loss": 0.9682,
+      "step": 304000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.452089501252562e-05,
+      "loss": 0.9685,
+      "step": 304500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.449547776295669e-05,
+      "loss": 0.9583,
+      "step": 305000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.447006051338778e-05,
+      "loss": 0.9745,
+      "step": 305500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.444464326381886e-05,
+      "loss": 0.9601,
+      "step": 306000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.441922601424993e-05,
+      "loss": 0.9656,
+      "step": 306500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.439380876468101e-05,
+      "loss": 0.98,
+      "step": 307000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.436839151511208e-05,
+      "loss": 0.9582,
+      "step": 307500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.434297426554316e-05,
+      "loss": 0.9644,
+      "step": 308000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.4317557015974236e-05,
+      "loss": 0.9402,
+      "step": 308500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4292139766405314e-05,
+      "loss": 0.9696,
+      "step": 309000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4266722516836386e-05,
+      "loss": 0.9727,
+      "step": 309500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4241305267267464e-05,
+      "loss": 0.9606,
+      "step": 310000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4215888017698536e-05,
+      "loss": 0.9508,
+      "step": 310500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.4190470768129614e-05,
+      "loss": 0.9421,
+      "step": 311000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.416505351856069e-05,
+      "loss": 0.9587,
+      "step": 311500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.413963626899177e-05,
+      "loss": 0.9409,
+      "step": 312000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.411421901942285e-05,
+      "loss": 0.9692,
+      "step": 312500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.408880176985392e-05,
+      "loss": 0.99,
+      "step": 313000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.4063384520285e-05,
+      "loss": 0.9441,
+      "step": 313500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.403796727071608e-05,
+      "loss": 0.969,
+      "step": 314000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.401255002114716e-05,
+      "loss": 0.9534,
+      "step": 314500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.398713277157823e-05,
+      "loss": 0.9752,
+      "step": 315000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.396171552200931e-05,
+      "loss": 0.9754,
+      "step": 315500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.393629827244038e-05,
+      "loss": 0.9676,
+      "step": 316000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.391088102287146e-05,
+      "loss": 0.9664,
+      "step": 316500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.3885463773302536e-05,
+      "loss": 0.9682,
+      "step": 317000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.3860046523733614e-05,
+      "loss": 0.9619,
+      "step": 317500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.3834629274164686e-05,
+      "loss": 0.9538,
+      "step": 318000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.3809212024595764e-05,
+      "loss": 0.9627,
+      "step": 318500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.378379477502684e-05,
+      "loss": 0.962,
+      "step": 319000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.375837752545792e-05,
+      "loss": 0.9782,
+      "step": 319500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.3732960275889e-05,
+      "loss": 0.9505,
+      "step": 320000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.370754302632007e-05,
+      "loss": 0.973,
+      "step": 320500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.368212577675115e-05,
+      "loss": 0.9497,
+      "step": 321000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.365670852718222e-05,
+      "loss": 0.9871,
+      "step": 321500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.36312912776133e-05,
+      "loss": 0.9439,
+      "step": 322000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.360587402804438e-05,
+      "loss": 0.9622,
+      "step": 322500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.358045677847546e-05,
+      "loss": 0.9762,
+      "step": 323000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.355503952890653e-05,
+      "loss": 0.9684,
+      "step": 323500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.352962227933761e-05,
+      "loss": 0.9594,
+      "step": 324000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.350420502976868e-05,
+      "loss": 0.9637,
+      "step": 324500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.3478787780199764e-05,
+      "loss": 0.9612,
+      "step": 325000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.345337053063084e-05,
+      "loss": 0.9428,
+      "step": 325500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3427953281061914e-05,
+      "loss": 0.9719,
+      "step": 326000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.340253603149299e-05,
+      "loss": 0.9764,
+      "step": 326500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.3377118781924064e-05,
+      "loss": 0.9764,
+      "step": 327000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.335170153235514e-05,
+      "loss": 0.9274,
+      "step": 327500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.332628428278622e-05,
+      "loss": 0.9725,
+      "step": 328000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.33008670332173e-05,
+      "loss": 0.9344,
+      "step": 328500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.327544978364837e-05,
+      "loss": 0.9546,
+      "step": 329000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.325003253407945e-05,
+      "loss": 0.9415,
+      "step": 329500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.322461528451052e-05,
+      "loss": 0.9724,
+      "step": 330000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.31991980349416e-05,
+      "loss": 0.9458,
+      "step": 330500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.317378078537268e-05,
+      "loss": 0.9698,
+      "step": 331000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.314836353580376e-05,
+      "loss": 0.9765,
+      "step": 331500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3122946286234835e-05,
+      "loss": 0.9579,
+      "step": 332000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.309752903666591e-05,
+      "loss": 0.9831,
+      "step": 332500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.3072111787096985e-05,
+      "loss": 0.9441,
+      "step": 333000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.3046694537528064e-05,
+      "loss": 0.9859,
+      "step": 333500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.302127728795914e-05,
+      "loss": 0.9519,
+      "step": 334000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.2995860038390214e-05,
+      "loss": 0.9389,
+      "step": 334500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.297044278882129e-05,
+      "loss": 0.9466,
+      "step": 335000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.2945025539252364e-05,
+      "loss": 0.9555,
+      "step": 335500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.291960828968344e-05,
+      "loss": 0.9674,
+      "step": 336000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.289419104011452e-05,
+      "loss": 0.9294,
+      "step": 336500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.28687737905456e-05,
+      "loss": 0.9363,
+      "step": 337000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.284335654097667e-05,
+      "loss": 0.9513,
+      "step": 337500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.281793929140775e-05,
+      "loss": 0.9646,
+      "step": 338000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.279252204183883e-05,
+      "loss": 0.968,
+      "step": 338500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.276710479226991e-05,
+      "loss": 0.944,
+      "step": 339000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.2741687542700985e-05,
+      "loss": 0.9634,
+      "step": 339500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.271627029313206e-05,
+      "loss": 0.9777,
+      "step": 340000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.2690853043563135e-05,
+      "loss": 0.9467,
+      "step": 340500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.266543579399421e-05,
+      "loss": 0.9379,
+      "step": 341000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.2640018544425285e-05,
+      "loss": 0.9388,
+      "step": 341500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.2614601294856364e-05,
+      "loss": 0.9602,
+      "step": 342000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.258918404528744e-05,
+      "loss": 0.9573,
+      "step": 342500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.2563766795718514e-05,
+      "loss": 0.9554,
+      "step": 343000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.253834954614959e-05,
+      "loss": 0.9599,
+      "step": 343500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.2512932296580664e-05,
+      "loss": 0.9615,
+      "step": 344000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.248751504701175e-05,
+      "loss": 0.9359,
+      "step": 344500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.246209779744283e-05,
+      "loss": 0.9473,
+      "step": 345000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.24366805478739e-05,
+      "loss": 0.9387,
+      "step": 345500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.241126329830498e-05,
+      "loss": 0.9731,
+      "step": 346000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.238584604873605e-05,
+      "loss": 0.9508,
+      "step": 346500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.236042879916713e-05,
+      "loss": 0.9817,
+      "step": 347000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.233501154959821e-05,
+      "loss": 0.9768,
+      "step": 347500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2309594300029285e-05,
+      "loss": 0.9476,
+      "step": 348000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.228417705046036e-05,
+      "loss": 0.9691,
+      "step": 348500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2258759800891435e-05,
+      "loss": 0.9447,
+      "step": 349000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.223334255132251e-05,
+      "loss": 0.9565,
+      "step": 349500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.2207925301753585e-05,
+      "loss": 0.9519,
+      "step": 350000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2182508052184664e-05,
+      "loss": 0.9388,
+      "step": 350500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.215709080261574e-05,
+      "loss": 0.9452,
+      "step": 351000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.213167355304682e-05,
+      "loss": 0.9322,
+      "step": 351500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.210625630347789e-05,
+      "loss": 0.9836,
+      "step": 352000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.208083905390897e-05,
+      "loss": 0.953,
+      "step": 352500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.205542180434005e-05,
+      "loss": 0.9438,
+      "step": 353000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.203000455477113e-05,
+      "loss": 0.968,
+      "step": 353500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.20045873052022e-05,
+      "loss": 0.9678,
+      "step": 354000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.197917005563328e-05,
+      "loss": 0.9638,
+      "step": 354500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.195375280606435e-05,
+      "loss": 0.925,
+      "step": 355000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.192833555649543e-05,
+      "loss": 0.9518,
+      "step": 355500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.190291830692651e-05,
+      "loss": 0.9366,
+      "step": 356000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.1877501057357585e-05,
+      "loss": 0.9562,
+      "step": 356500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.185208380778866e-05,
+      "loss": 0.9533,
+      "step": 357000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.1826666558219735e-05,
+      "loss": 0.9533,
+      "step": 357500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.1801249308650814e-05,
+      "loss": 0.9427,
+      "step": 358000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.177583205908189e-05,
+      "loss": 0.9498,
+      "step": 358500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.175041480951297e-05,
+      "loss": 0.9681,
+      "step": 359000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.172499755994404e-05,
+      "loss": 0.9256,
+      "step": 359500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.169958031037512e-05,
+      "loss": 0.9384,
+      "step": 360000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.167416306080619e-05,
+      "loss": 0.9359,
+      "step": 360500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.164874581123727e-05,
+      "loss": 0.9201,
+      "step": 361000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.162332856166835e-05,
+      "loss": 0.9155,
+      "step": 361500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.159791131209943e-05,
+      "loss": 0.93,
+      "step": 362000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.15724940625305e-05,
+      "loss": 0.9417,
+      "step": 362500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.154707681296158e-05,
+      "loss": 0.9378,
+      "step": 363000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.152165956339265e-05,
+      "loss": 0.9603,
+      "step": 363500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1496242313823735e-05,
+      "loss": 0.9419,
+      "step": 364000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1470825064254814e-05,
+      "loss": 0.9434,
+      "step": 364500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.1445407814685885e-05,
+      "loss": 0.9592,
+      "step": 365000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1419990565116964e-05,
+      "loss": 0.9467,
+      "step": 365500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1394573315548035e-05,
+      "loss": 0.9213,
+      "step": 366000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1369156065979114e-05,
+      "loss": 0.9379,
+      "step": 366500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.134373881641019e-05,
+      "loss": 0.9458,
+      "step": 367000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.131832156684127e-05,
+      "loss": 0.9452,
+      "step": 367500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.129290431727234e-05,
+      "loss": 0.954,
+      "step": 368000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.126748706770342e-05,
+      "loss": 0.915,
+      "step": 368500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.124206981813449e-05,
+      "loss": 0.9357,
+      "step": 369000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.121665256856557e-05,
+      "loss": 0.9628,
+      "step": 369500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.119123531899665e-05,
+      "loss": 0.9303,
+      "step": 370000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.116581806942773e-05,
+      "loss": 0.9509,
+      "step": 370500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1140400819858806e-05,
+      "loss": 0.9311,
+      "step": 371000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.111498357028988e-05,
+      "loss": 0.9415,
+      "step": 371500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1089566320720957e-05,
+      "loss": 0.9479,
+      "step": 372000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.1064149071152035e-05,
+      "loss": 0.9461,
+      "step": 372500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1038731821583114e-05,
+      "loss": 0.9196,
+      "step": 373000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.1013314572014185e-05,
+      "loss": 0.945,
+      "step": 373500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.0987897322445264e-05,
+      "loss": 0.9388,
+      "step": 374000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.0962480072876335e-05,
+      "loss": 0.9407,
+      "step": 374500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.0937062823307414e-05,
+      "loss": 0.9476,
+      "step": 375000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.091164557373849e-05,
+      "loss": 0.9381,
+      "step": 375500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.088622832416957e-05,
+      "loss": 0.9199,
+      "step": 376000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.086081107460064e-05,
+      "loss": 0.9425,
+      "step": 376500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.083539382503172e-05,
+      "loss": 0.9514,
+      "step": 377000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.08099765754628e-05,
+      "loss": 0.934,
+      "step": 377500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.078455932589388e-05,
+      "loss": 0.9539,
+      "step": 378000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0759142076324956e-05,
+      "loss": 0.9609,
+      "step": 378500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.073372482675603e-05,
+      "loss": 0.94,
+      "step": 379000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.0708307577187106e-05,
+      "loss": 0.9565,
+      "step": 379500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.068289032761818e-05,
+      "loss": 0.9328,
+      "step": 380000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0657473078049257e-05,
+      "loss": 0.9179,
+      "step": 380500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0632055828480335e-05,
+      "loss": 0.9669,
+      "step": 381000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0606638578911414e-05,
+      "loss": 0.9322,
+      "step": 381500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.0581221329342485e-05,
+      "loss": 0.9484,
+      "step": 382000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0555804079773564e-05,
+      "loss": 0.971,
+      "step": 382500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0530386830204635e-05,
+      "loss": 0.94,
+      "step": 383000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0504969580635717e-05,
+      "loss": 0.9083,
+      "step": 383500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.0479552331066796e-05,
+      "loss": 0.95,
+      "step": 384000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.045413508149787e-05,
+      "loss": 0.9102,
+      "step": 384500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.042871783192895e-05,
+      "loss": 0.9302,
+      "step": 385000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.040330058236002e-05,
+      "loss": 0.9394,
+      "step": 385500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0377883332791103e-05,
+      "loss": 0.9425,
+      "step": 386000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0352466083222174e-05,
+      "loss": 0.9223,
+      "step": 386500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.0327048833653253e-05,
+      "loss": 0.9579,
+      "step": 387000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0301631584084328e-05,
+      "loss": 0.9317,
+      "step": 387500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.0276214334515406e-05,
+      "loss": 0.9324,
+      "step": 388000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.025079708494648e-05,
+      "loss": 0.9453,
+      "step": 388500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.022537983537756e-05,
+      "loss": 0.9205,
+      "step": 389000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.019996258580863e-05,
+      "loss": 0.9175,
+      "step": 389500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0174545336239714e-05,
+      "loss": 0.9338,
+      "step": 390000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0149128086670792e-05,
+      "loss": 0.9728,
+      "step": 390500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0123710837101864e-05,
+      "loss": 0.9429,
+      "step": 391000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0098293587532946e-05,
+      "loss": 0.9326,
+      "step": 391500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.0072876337964017e-05,
+      "loss": 0.9476,
+      "step": 392000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.0047459088395096e-05,
+      "loss": 0.9271,
+      "step": 392500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.002204183882617e-05,
+      "loss": 0.9241,
+      "step": 393000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.999662458925725e-05,
+      "loss": 0.9156,
+      "step": 393500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9971207339688324e-05,
+      "loss": 0.9234,
+      "step": 394000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.9945790090119403e-05,
+      "loss": 0.9183,
+      "step": 394500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9920372840550474e-05,
+      "loss": 0.9171,
+      "step": 395000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9894955590981556e-05,
+      "loss": 0.9431,
+      "step": 395500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9869538341412628e-05,
+      "loss": 0.9258,
+      "step": 396000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9844121091843706e-05,
+      "loss": 0.9319,
+      "step": 396500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.9818703842274788e-05,
+      "loss": 0.9384,
+      "step": 397000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.979328659270586e-05,
+      "loss": 0.933,
+      "step": 397500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.976786934313694e-05,
+      "loss": 0.9315,
+      "step": 398000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9742452093568014e-05,
+      "loss": 0.9354,
+      "step": 398500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9717034843999092e-05,
+      "loss": 0.9429,
+      "step": 399000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.9691617594430167e-05,
+      "loss": 0.9539,
+      "step": 399500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9666200344861246e-05,
+      "loss": 0.9367,
+      "step": 400000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9640783095292317e-05,
+      "loss": 0.9279,
+      "step": 400500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.9615365845723396e-05,
+      "loss": 0.9207,
+      "step": 401000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.958994859615447e-05,
+      "loss": 0.9312,
+      "step": 401500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.956453134658555e-05,
+      "loss": 0.9303,
+      "step": 402000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9539114097016624e-05,
+      "loss": 0.9222,
+      "step": 402500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9513696847447703e-05,
+      "loss": 0.9325,
+      "step": 403000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.948827959787878e-05,
+      "loss": 0.8994,
+      "step": 403500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.9462862348309856e-05,
+      "loss": 0.9334,
+      "step": 404000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9437445098740935e-05,
+      "loss": 0.9132,
+      "step": 404500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.9412027849172006e-05,
+      "loss": 0.905,
+      "step": 405000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.938661059960309e-05,
+      "loss": 0.9003,
+      "step": 405500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.936119335003416e-05,
+      "loss": 0.9128,
+      "step": 406000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.933577610046524e-05,
+      "loss": 0.9223,
+      "step": 406500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9310358850896314e-05,
+      "loss": 0.936,
+      "step": 407000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9284941601327392e-05,
+      "loss": 0.923,
+      "step": 407500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9259524351758467e-05,
+      "loss": 0.952,
+      "step": 408000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9234107102189546e-05,
+      "loss": 0.9129,
+      "step": 408500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.9208689852620617e-05,
+      "loss": 0.9554,
+      "step": 409000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.91832726030517e-05,
+      "loss": 0.9257,
+      "step": 409500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9157855353482778e-05,
+      "loss": 0.9218,
+      "step": 410000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.913243810391385e-05,
+      "loss": 0.9262,
+      "step": 410500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.910702085434493e-05,
+      "loss": 0.9485,
+      "step": 411000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.9081603604776003e-05,
+      "loss": 0.9358,
+      "step": 411500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.905618635520708e-05,
+      "loss": 0.9504,
+      "step": 412000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9030769105638156e-05,
+      "loss": 0.9086,
+      "step": 412500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.9005351856069235e-05,
+      "loss": 0.9008,
+      "step": 413000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.897993460650031e-05,
+      "loss": 0.9314,
+      "step": 413500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.895451735693139e-05,
+      "loss": 0.9303,
+      "step": 414000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.892910010736246e-05,
+      "loss": 0.9375,
+      "step": 414500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.8903682857793542e-05,
+      "loss": 0.9449,
+      "step": 415000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.8878265608224614e-05,
+      "loss": 0.923,
+      "step": 415500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.8852848358655692e-05,
+      "loss": 0.9328,
+      "step": 416000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.882743110908677e-05,
+      "loss": 0.9258,
+      "step": 416500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8802013859517846e-05,
+      "loss": 0.9337,
+      "step": 417000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8776596609948924e-05,
+      "loss": 0.9398,
+      "step": 417500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.875117936038e-05,
+      "loss": 0.9093,
+      "step": 418000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8725762110811078e-05,
+      "loss": 0.9255,
+      "step": 418500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8700344861242153e-05,
+      "loss": 0.9257,
+      "step": 419000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.867492761167323e-05,
+      "loss": 0.9025,
+      "step": 419500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8649510362104303e-05,
+      "loss": 0.9627,
+      "step": 420000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.862409311253538e-05,
+      "loss": 0.9314,
+      "step": 420500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8598675862966456e-05,
+      "loss": 0.9289,
+      "step": 421000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8573258613397535e-05,
+      "loss": 0.8998,
+      "step": 421500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.854784136382861e-05,
+      "loss": 0.9139,
+      "step": 422000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.852242411425969e-05,
+      "loss": 0.935,
+      "step": 422500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8497006864690767e-05,
+      "loss": 0.9168,
+      "step": 423000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.8471589615121842e-05,
+      "loss": 0.9348,
+      "step": 423500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.844617236555292e-05,
+      "loss": 0.9167,
+      "step": 424000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8420755115983992e-05,
+      "loss": 0.9165,
+      "step": 424500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8395337866415074e-05,
+      "loss": 0.9122,
+      "step": 425000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8369920616846146e-05,
+      "loss": 0.923,
+      "step": 425500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.8344503367277224e-05,
+      "loss": 0.9316,
+      "step": 426000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.83190861177083e-05,
+      "loss": 0.9236,
+      "step": 426500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8293668868139378e-05,
+      "loss": 0.9402,
+      "step": 427000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8268251618570453e-05,
+      "loss": 0.9165,
+      "step": 427500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.824283436900153e-05,
+      "loss": 0.9327,
+      "step": 428000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8217417119432603e-05,
+      "loss": 0.9045,
+      "step": 428500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.8191999869863685e-05,
+      "loss": 0.9058,
+      "step": 429000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8166582620294763e-05,
+      "loss": 0.9101,
+      "step": 429500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8141165370725835e-05,
+      "loss": 0.9085,
+      "step": 430000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8115748121156917e-05,
+      "loss": 0.9406,
+      "step": 430500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.809033087158799e-05,
+      "loss": 0.9153,
+      "step": 431000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.8064913622019067e-05,
+      "loss": 0.9242,
+      "step": 431500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.8039496372450142e-05,
+      "loss": 0.9094,
+      "step": 432000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.801407912288122e-05,
+      "loss": 0.9165,
+      "step": 432500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7988661873312295e-05,
+      "loss": 0.9348,
+      "step": 433000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7963244623743374e-05,
+      "loss": 0.9178,
+      "step": 433500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.7937827374174446e-05,
+      "loss": 0.8945,
+      "step": 434000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7912410124605527e-05,
+      "loss": 0.9214,
+      "step": 434500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.78869928750366e-05,
+      "loss": 0.9281,
+      "step": 435000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7861575625467678e-05,
+      "loss": 0.9263,
+      "step": 435500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.7836158375898756e-05,
+      "loss": 0.919,
+      "step": 436000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.781074112632983e-05,
+      "loss": 0.9135,
+      "step": 436500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.778532387676091e-05,
+      "loss": 0.9166,
+      "step": 437000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7759906627191985e-05,
+      "loss": 0.9354,
+      "step": 437500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7734489377623063e-05,
+      "loss": 0.9432,
+      "step": 438000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.7709072128054138e-05,
+      "loss": 0.9276,
+      "step": 438500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7683654878485217e-05,
+      "loss": 0.9074,
+      "step": 439000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.765823762891629e-05,
+      "loss": 0.9196,
+      "step": 439500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7632820379347367e-05,
+      "loss": 0.9283,
+      "step": 440000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.7607403129778442e-05,
+      "loss": 0.9229,
+      "step": 440500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.758198588020952e-05,
+      "loss": 0.9176,
+      "step": 441000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7556568630640595e-05,
+      "loss": 0.9127,
+      "step": 441500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7531151381071674e-05,
+      "loss": 0.8947,
+      "step": 442000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7505734131502752e-05,
+      "loss": 0.9159,
+      "step": 442500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7480316881933827e-05,
+      "loss": 0.9098,
+      "step": 443000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.7454899632364906e-05,
+      "loss": 0.9496,
+      "step": 443500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7429482382795978e-05,
+      "loss": 0.8969,
+      "step": 444000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.740406513322706e-05,
+      "loss": 0.9333,
+      "step": 444500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.737864788365813e-05,
+      "loss": 0.926,
+      "step": 445000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.735323063408921e-05,
+      "loss": 0.8983,
+      "step": 445500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7327813384520285e-05,
+      "loss": 0.92,
+      "step": 446000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7302396134951363e-05,
+      "loss": 0.8873,
+      "step": 446500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7276978885382438e-05,
+      "loss": 0.9294,
+      "step": 447000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7251561635813517e-05,
+      "loss": 0.9436,
+      "step": 447500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.722614438624459e-05,
+      "loss": 0.9044,
+      "step": 448000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.720072713667567e-05,
+      "loss": 0.9145,
+      "step": 448500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.717530988710675e-05,
+      "loss": 0.9133,
+      "step": 449000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.714989263753782e-05,
+      "loss": 0.9309,
+      "step": 449500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7124475387968902e-05,
+      "loss": 0.8809,
+      "step": 450000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7099058138399974e-05,
+      "loss": 0.9263,
+      "step": 450500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.7073640888831052e-05,
+      "loss": 0.9157,
+      "step": 451000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7048223639262127e-05,
+      "loss": 0.9016,
+      "step": 451500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.7022806389693206e-05,
+      "loss": 0.9095,
+      "step": 452000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.699738914012428e-05,
+      "loss": 0.917,
+      "step": 452500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.697197189055536e-05,
+      "loss": 0.9366,
+      "step": 453000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.694655464098643e-05,
+      "loss": 0.9035,
+      "step": 453500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6921137391417513e-05,
+      "loss": 0.9298,
+      "step": 454000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.689572014184859e-05,
+      "loss": 0.9219,
+      "step": 454500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6870302892279663e-05,
+      "loss": 0.9027,
+      "step": 455000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.684488564271074e-05,
+      "loss": 0.9108,
+      "step": 455500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.6819468393141817e-05,
+      "loss": 0.944,
+      "step": 456000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6794051143572895e-05,
+      "loss": 0.8896,
+      "step": 456500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.676863389400397e-05,
+      "loss": 0.9308,
+      "step": 457000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.674321664443505e-05,
+      "loss": 0.9071,
+      "step": 457500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.671779939486612e-05,
+      "loss": 0.897,
+      "step": 458000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6692382145297202e-05,
+      "loss": 0.926,
+      "step": 458500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6666964895728274e-05,
+      "loss": 0.9054,
+      "step": 459000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6641547646159352e-05,
+      "loss": 0.9199,
+      "step": 459500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6616130396590427e-05,
+      "loss": 0.8835,
+      "step": 460000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6590713147021506e-05,
+      "loss": 0.904,
+      "step": 460500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6565295897452584e-05,
+      "loss": 0.9012,
+      "step": 461000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.653987864788366e-05,
+      "loss": 0.9259,
+      "step": 461500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6514461398314738e-05,
+      "loss": 0.9077,
+      "step": 462000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6489044148745813e-05,
+      "loss": 0.9027,
+      "step": 462500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.646362689917689e-05,
+      "loss": 0.9127,
+      "step": 463000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.6438209649607963e-05,
+      "loss": 0.9227,
+      "step": 463500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6412792400039045e-05,
+      "loss": 0.9295,
+      "step": 464000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6387375150470117e-05,
+      "loss": 0.8926,
+      "step": 464500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.6361957900901195e-05,
+      "loss": 0.9059,
+      "step": 465000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.633654065133227e-05,
+      "loss": 0.9317,
+      "step": 465500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.631112340176335e-05,
+      "loss": 0.9224,
+      "step": 466000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6285706152194424e-05,
+      "loss": 0.9168,
+      "step": 466500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6260288902625502e-05,
+      "loss": 0.9187,
+      "step": 467000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.623487165305658e-05,
+      "loss": 0.8928,
+      "step": 467500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.6209454403487656e-05,
+      "loss": 0.9036,
+      "step": 468000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6184037153918734e-05,
+      "loss": 0.9039,
+      "step": 468500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6158619904349806e-05,
+      "loss": 0.918,
+      "step": 469000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6133202654780888e-05,
+      "loss": 0.905,
+      "step": 469500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.610778540521196e-05,
+      "loss": 0.9,
+      "step": 470000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6082368155643038e-05,
+      "loss": 0.9201,
+      "step": 470500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6056950906074113e-05,
+      "loss": 0.9119,
+      "step": 471000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.603153365650519e-05,
+      "loss": 0.8895,
+      "step": 471500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.6006116406936267e-05,
+      "loss": 0.9023,
+      "step": 472000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5980699157367345e-05,
+      "loss": 0.9262,
+      "step": 472500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.5955281907798417e-05,
+      "loss": 0.9106,
+      "step": 473000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.59298646582295e-05,
+      "loss": 0.917,
+      "step": 473500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.5904447408660577e-05,
+      "loss": 0.8921,
+      "step": 474000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.587903015909165e-05,
+      "loss": 0.8966,
+      "step": 474500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.5853612909522727e-05,
+      "loss": 0.9263,
+      "step": 475000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.5828195659953802e-05,
+      "loss": 0.8993,
+      "step": 475500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.580277841038488e-05,
+      "loss": 0.9014,
+      "step": 476000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5777361160815956e-05,
+      "loss": 0.942,
+      "step": 476500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5751943911247034e-05,
+      "loss": 0.8848,
+      "step": 477000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5726526661678106e-05,
+      "loss": 0.8987,
+      "step": 477500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.5701109412109188e-05,
+      "loss": 0.945,
+      "step": 478000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.567569216254026e-05,
+      "loss": 0.8911,
+      "step": 478500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5650274912971338e-05,
+      "loss": 0.9125,
+      "step": 479000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.5624857663402413e-05,
+      "loss": 0.8895,
+      "step": 479500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.559944041383349e-05,
+      "loss": 0.9034,
+      "step": 480000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.557402316426457e-05,
+      "loss": 0.9032,
+      "step": 480500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5548605914695645e-05,
+      "loss": 0.8868,
+      "step": 481000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5523188665126723e-05,
+      "loss": 0.9331,
+      "step": 481500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.54977714155578e-05,
+      "loss": 0.8887,
+      "step": 482000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5472354165988877e-05,
+      "loss": 0.8952,
+      "step": 482500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.544693691641995e-05,
+      "loss": 0.9393,
+      "step": 483000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.542151966685103e-05,
+      "loss": 0.9066,
+      "step": 483500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5396102417282102e-05,
+      "loss": 0.8819,
+      "step": 484000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.537068516771318e-05,
+      "loss": 0.9045,
+      "step": 484500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5345267918144256e-05,
+      "loss": 0.9049,
+      "step": 485000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.5319850668575334e-05,
+      "loss": 0.9111,
+      "step": 485500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.529443341900641e-05,
+      "loss": 0.9053,
+      "step": 486000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5269016169437488e-05,
+      "loss": 0.9122,
+      "step": 486500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5243598919868566e-05,
+      "loss": 0.9093,
+      "step": 487000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.521818167029964e-05,
+      "loss": 0.9216,
+      "step": 487500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.519276442073072e-05,
+      "loss": 0.8975,
+      "step": 488000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.516734717116179e-05,
+      "loss": 0.934,
+      "step": 488500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5141929921592873e-05,
+      "loss": 0.8897,
+      "step": 489000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5116512672023945e-05,
+      "loss": 0.9056,
+      "step": 489500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5091095422455023e-05,
+      "loss": 0.9146,
+      "step": 490000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.50656781728861e-05,
+      "loss": 0.886,
+      "step": 490500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5040260923317177e-05,
+      "loss": 0.9303,
+      "step": 491000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.5014843673748252e-05,
+      "loss": 0.9115,
+      "step": 491500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.498942642417933e-05,
+      "loss": 0.9044,
+      "step": 492000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.4964009174610406e-05,
+      "loss": 0.9064,
+      "step": 492500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.493859192504148e-05,
+      "loss": 0.8855,
+      "step": 493000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.491317467547256e-05,
+      "loss": 0.9246,
+      "step": 493500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4887757425903634e-05,
+      "loss": 0.9049,
+      "step": 494000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4862340176334713e-05,
+      "loss": 0.919,
+      "step": 494500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.4836922926765788e-05,
+      "loss": 0.9279,
+      "step": 495000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4811505677196863e-05,
+      "loss": 0.8967,
+      "step": 495500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.478608842762794e-05,
+      "loss": 0.9023,
+      "step": 496000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4760671178059016e-05,
+      "loss": 0.9118,
+      "step": 496500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4735253928490095e-05,
+      "loss": 0.8974,
+      "step": 497000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.4709836678921173e-05,
+      "loss": 0.8932,
+      "step": 497500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.468441942935225e-05,
+      "loss": 0.8928,
+      "step": 498000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4659002179783323e-05,
+      "loss": 0.9004,
+      "step": 498500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4633584930214402e-05,
+      "loss": 0.8938,
+      "step": 499000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4608167680645477e-05,
+      "loss": 0.8962,
+      "step": 499500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.4582750431076555e-05,
+      "loss": 0.9106,
+      "step": 500000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.455733318150763e-05,
+      "loss": 0.8882,
+      "step": 500500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4531915931938706e-05,
+      "loss": 0.8976,
+      "step": 501000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4506498682369784e-05,
+      "loss": 0.8805,
+      "step": 501500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.448108143280086e-05,
+      "loss": 0.8714,
+      "step": 502000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.4455664183231934e-05,
+      "loss": 0.9048,
+      "step": 502500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4430246933663013e-05,
+      "loss": 0.8891,
+      "step": 503000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.440482968409409e-05,
+      "loss": 0.9297,
+      "step": 503500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4379412434525166e-05,
+      "loss": 0.8957,
+      "step": 504000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.4353995184956245e-05,
+      "loss": 0.9294,
+      "step": 504500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.432857793538732e-05,
+      "loss": 0.9139,
+      "step": 505000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4303160685818395e-05,
+      "loss": 0.8913,
+      "step": 505500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4277743436249473e-05,
+      "loss": 0.9195,
+      "step": 506000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.425232618668055e-05,
+      "loss": 0.8833,
+      "step": 506500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4226908937111627e-05,
+      "loss": 0.911,
+      "step": 507000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.4201491687542702e-05,
+      "loss": 0.8757,
+      "step": 507500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4176074437973777e-05,
+      "loss": 0.8941,
+      "step": 508000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4150657188404855e-05,
+      "loss": 0.9034,
+      "step": 508500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.412523993883593e-05,
+      "loss": 0.9018,
+      "step": 509000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4099822689267006e-05,
+      "loss": 0.8931,
+      "step": 509500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.4074405439698087e-05,
+      "loss": 0.8842,
+      "step": 510000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4048988190129163e-05,
+      "loss": 0.8968,
+      "step": 510500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.4023570940560238e-05,
+      "loss": 0.906,
+      "step": 511000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3998153690991316e-05,
+      "loss": 0.906,
+      "step": 511500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.397273644142239e-05,
+      "loss": 0.9217,
+      "step": 512000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.3947319191853466e-05,
+      "loss": 0.8863,
+      "step": 512500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3921901942284545e-05,
+      "loss": 0.8877,
+      "step": 513000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.389648469271562e-05,
+      "loss": 0.9131,
+      "step": 513500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3871067443146698e-05,
+      "loss": 0.9141,
+      "step": 514000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.3845650193577773e-05,
+      "loss": 0.8853,
+      "step": 514500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.382023294400885e-05,
+      "loss": 0.8876,
+      "step": 515000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3794815694439927e-05,
+      "loss": 0.9033,
+      "step": 515500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3769398444871002e-05,
+      "loss": 0.8836,
+      "step": 516000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.374398119530208e-05,
+      "loss": 0.8901,
+      "step": 516500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.371856394573316e-05,
+      "loss": 0.8639,
+      "step": 517000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.3693146696164234e-05,
+      "loss": 0.902,
+      "step": 517500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.366772944659531e-05,
+      "loss": 0.8842,
+      "step": 518000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3642312197026387e-05,
+      "loss": 0.8917,
+      "step": 518500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3616894947457463e-05,
+      "loss": 0.8707,
+      "step": 519000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.359147769788854e-05,
+      "loss": 0.8959,
+      "step": 519500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.3566060448319616e-05,
+      "loss": 0.874,
+      "step": 520000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.354064319875069e-05,
+      "loss": 0.8986,
+      "step": 520500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.351522594918177e-05,
+      "loss": 0.8915,
+      "step": 521000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3489808699612845e-05,
+      "loss": 0.9105,
+      "step": 521500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.346439145004392e-05,
+      "loss": 0.8883,
+      "step": 522000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3438974200474998e-05,
+      "loss": 0.8779,
+      "step": 522500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3413556950906077e-05,
+      "loss": 0.9073,
+      "step": 523000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3388139701337152e-05,
+      "loss": 0.8639,
+      "step": 523500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.336272245176823e-05,
+      "loss": 0.8795,
+      "step": 524000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3337305202199305e-05,
+      "loss": 0.8971,
+      "step": 524500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.331188795263038e-05,
+      "loss": 0.8783,
+      "step": 525000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.328647070306146e-05,
+      "loss": 0.9056,
+      "step": 525500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3261053453492534e-05,
+      "loss": 0.8759,
+      "step": 526000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3235636203923612e-05,
+      "loss": 0.8968,
+      "step": 526500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3210218954354687e-05,
+      "loss": 0.8828,
+      "step": 527000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3184801704785763e-05,
+      "loss": 0.8911,
+      "step": 527500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.315938445521684e-05,
+      "loss": 0.8905,
+      "step": 528000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3133967205647916e-05,
+      "loss": 0.8985,
+      "step": 528500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.310854995607899e-05,
+      "loss": 0.8843,
+      "step": 529000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3083132706510073e-05,
+      "loss": 0.9017,
+      "step": 529500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3057715456941148e-05,
+      "loss": 0.8776,
+      "step": 530000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3032298207372223e-05,
+      "loss": 0.8966,
+      "step": 530500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.30068809578033e-05,
+      "loss": 0.8849,
+      "step": 531000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.2981463708234377e-05,
+      "loss": 0.901,
+      "step": 531500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.2956046458665452e-05,
+      "loss": 0.9011,
+      "step": 532000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.293062920909653e-05,
+      "loss": 0.8715,
+      "step": 532500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2905211959527605e-05,
+      "loss": 0.8939,
+      "step": 533000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2879794709958684e-05,
+      "loss": 0.8877,
+      "step": 533500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.285437746038976e-05,
+      "loss": 0.8861,
+      "step": 534000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2828960210820834e-05,
+      "loss": 0.8971,
+      "step": 534500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2803542961251912e-05,
+      "loss": 0.9101,
+      "step": 535000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2778125711682988e-05,
+      "loss": 0.8779,
+      "step": 535500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2752708462114066e-05,
+      "loss": 0.8575,
+      "step": 536000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2727291212545144e-05,
+      "loss": 0.9062,
+      "step": 536500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.270187396297622e-05,
+      "loss": 0.9015,
+      "step": 537000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2676456713407295e-05,
+      "loss": 0.8742,
+      "step": 537500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2651039463838373e-05,
+      "loss": 0.8795,
+      "step": 538000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2625622214269448e-05,
+      "loss": 0.8854,
+      "step": 538500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.2600204964700523e-05,
+      "loss": 0.9019,
+      "step": 539000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.25747877151316e-05,
+      "loss": 0.8807,
+      "step": 539500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2549370465562677e-05,
+      "loss": 0.8893,
+      "step": 540000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2523953215993755e-05,
+      "loss": 0.8972,
+      "step": 540500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.249853596642483e-05,
+      "loss": 0.9028,
+      "step": 541000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2473118716855905e-05,
+      "loss": 0.8806,
+      "step": 541500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.2447701467286984e-05,
+      "loss": 0.8957,
+      "step": 542000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2422284217718062e-05,
+      "loss": 0.8769,
+      "step": 542500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2396866968149137e-05,
+      "loss": 0.9111,
+      "step": 543000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2371449718580216e-05,
+      "loss": 0.9003,
+      "step": 543500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.234603246901129e-05,
+      "loss": 0.8917,
+      "step": 544000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.2320615219442366e-05,
+      "loss": 0.9053,
+      "step": 544500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2295197969873444e-05,
+      "loss": 0.8928,
+      "step": 545000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.226978072030452e-05,
+      "loss": 0.8857,
+      "step": 545500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2244363470735598e-05,
+      "loss": 0.8945,
+      "step": 546000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2218946221166673e-05,
+      "loss": 0.858,
+      "step": 546500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.2193528971597748e-05,
+      "loss": 0.855,
+      "step": 547000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2168111722028827e-05,
+      "loss": 0.9023,
+      "step": 547500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.21426944724599e-05,
+      "loss": 0.8932,
+      "step": 548000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2117277222890977e-05,
+      "loss": 0.8949,
+      "step": 548500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.209185997332206e-05,
+      "loss": 0.9008,
+      "step": 549000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.2066442723753134e-05,
+      "loss": 0.9099,
+      "step": 549500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.204102547418421e-05,
+      "loss": 0.8956,
+      "step": 550000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.2015608224615287e-05,
+      "loss": 0.9098,
+      "step": 550500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1990190975046362e-05,
+      "loss": 0.8761,
+      "step": 551000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1964773725477437e-05,
+      "loss": 0.8627,
+      "step": 551500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1939356475908516e-05,
+      "loss": 0.8824,
+      "step": 552000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.191393922633959e-05,
+      "loss": 0.882,
+      "step": 552500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.188852197677067e-05,
+      "loss": 0.8921,
+      "step": 553000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1863104727201744e-05,
+      "loss": 0.8978,
+      "step": 553500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.183768747763282e-05,
+      "loss": 0.882,
+      "step": 554000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1812270228063898e-05,
+      "loss": 0.8986,
+      "step": 554500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1786852978494973e-05,
+      "loss": 0.8642,
+      "step": 555000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.176143572892605e-05,
+      "loss": 0.8914,
+      "step": 555500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.173601847935713e-05,
+      "loss": 0.886,
+      "step": 556000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1710601229788205e-05,
+      "loss": 0.8772,
+      "step": 556500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.168518398021928e-05,
+      "loss": 0.8976,
+      "step": 557000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.165976673065036e-05,
+      "loss": 0.887,
+      "step": 557500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1634349481081434e-05,
+      "loss": 0.9007,
+      "step": 558000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.160893223151251e-05,
+      "loss": 0.8674,
+      "step": 558500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1583514981943587e-05,
+      "loss": 0.9017,
+      "step": 559000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1558097732374662e-05,
+      "loss": 0.8915,
+      "step": 559500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.153268048280574e-05,
+      "loss": 0.8915,
+      "step": 560000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1507263233236816e-05,
+      "loss": 0.8795,
+      "step": 560500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.148184598366789e-05,
+      "loss": 0.8768,
+      "step": 561000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.145642873409897e-05,
+      "loss": 0.898,
+      "step": 561500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.1431011484530048e-05,
+      "loss": 0.8629,
+      "step": 562000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.1405594234961123e-05,
+      "loss": 0.865,
+      "step": 562500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.13801769853922e-05,
+      "loss": 0.8776,
+      "step": 563000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.1354759735823276e-05,
+      "loss": 0.8695,
+      "step": 563500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.132934248625435e-05,
+      "loss": 0.8835,
+      "step": 564000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.130392523668543e-05,
+      "loss": 0.9007,
+      "step": 564500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.1278507987116505e-05,
+      "loss": 0.8711,
+      "step": 565000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.1253090737547584e-05,
+      "loss": 0.898,
+      "step": 565500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.122767348797866e-05,
+      "loss": 0.8901,
+      "step": 566000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.1202256238409734e-05,
+      "loss": 0.8925,
+      "step": 566500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1176838988840812e-05,
+      "loss": 0.8671,
+      "step": 567000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1151421739271887e-05,
+      "loss": 0.8802,
+      "step": 567500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1126004489702962e-05,
+      "loss": 0.8957,
+      "step": 568000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1100587240134044e-05,
+      "loss": 0.8787,
+      "step": 568500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.107516999056512e-05,
+      "loss": 0.9005,
+      "step": 569000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1049752740996194e-05,
+      "loss": 0.8795,
+      "step": 569500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1024335491427273e-05,
+      "loss": 0.8692,
+      "step": 570000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0998918241858348e-05,
+      "loss": 0.8882,
+      "step": 570500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0973500992289423e-05,
+      "loss": 0.9032,
+      "step": 571000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.09480837427205e-05,
+      "loss": 0.8773,
+      "step": 571500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0922666493151576e-05,
+      "loss": 0.8646,
+      "step": 572000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0897249243582655e-05,
+      "loss": 0.8752,
+      "step": 572500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.087183199401373e-05,
+      "loss": 0.8827,
+      "step": 573000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0846414744444805e-05,
+      "loss": 0.8925,
+      "step": 573500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0820997494875884e-05,
+      "loss": 0.8724,
+      "step": 574000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.079558024530696e-05,
+      "loss": 0.8747,
+      "step": 574500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0770162995738037e-05,
+      "loss": 0.8968,
+      "step": 575000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0744745746169116e-05,
+      "loss": 0.8906,
+      "step": 575500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.071932849660019e-05,
+      "loss": 0.9067,
+      "step": 576000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0693911247031266e-05,
+      "loss": 0.8772,
+      "step": 576500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0668493997462344e-05,
+      "loss": 0.8886,
+      "step": 577000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.064307674789342e-05,
+      "loss": 0.8772,
+      "step": 577500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0617659498324494e-05,
+      "loss": 0.875,
+      "step": 578000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0592242248755573e-05,
+      "loss": 0.8929,
+      "step": 578500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0566824999186648e-05,
+      "loss": 0.8869,
+      "step": 579000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0541407749617726e-05,
+      "loss": 0.889,
+      "step": 579500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.05159905000488e-05,
+      "loss": 0.912,
+      "step": 580000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0490573250479876e-05,
+      "loss": 0.8776,
+      "step": 580500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0465156000910955e-05,
+      "loss": 0.8672,
+      "step": 581000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0439738751342033e-05,
+      "loss": 0.8818,
+      "step": 581500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.041432150177311e-05,
+      "loss": 0.8525,
+      "step": 582000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0388904252204187e-05,
+      "loss": 0.879,
+      "step": 582500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0363487002635262e-05,
+      "loss": 0.8602,
+      "step": 583000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0338069753066337e-05,
+      "loss": 0.8917,
+      "step": 583500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0312652503497416e-05,
+      "loss": 0.8792,
+      "step": 584000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.028723525392849e-05,
+      "loss": 0.8923,
+      "step": 584500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0261818004359566e-05,
+      "loss": 0.8887,
+      "step": 585000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0236400754790644e-05,
+      "loss": 0.8828,
+      "step": 585500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.021098350522172e-05,
+      "loss": 0.8899,
+      "step": 586000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0185566255652798e-05,
+      "loss": 0.8782,
+      "step": 586500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0160149006083873e-05,
+      "loss": 0.8891,
+      "step": 587000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0134731756514948e-05,
+      "loss": 0.8872,
+      "step": 587500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.010931450694603e-05,
+      "loss": 0.883,
+      "step": 588000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0083897257377105e-05,
+      "loss": 0.8868,
+      "step": 588500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.005848000780818e-05,
+      "loss": 0.874,
+      "step": 589000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.003306275823926e-05,
+      "loss": 0.8716,
+      "step": 589500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0007645508670333e-05,
+      "loss": 0.873,
+      "step": 590000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.998222825910141e-05,
+      "loss": 0.8977,
+      "step": 590500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9956811009532487e-05,
+      "loss": 0.8912,
+      "step": 591000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9931393759963562e-05,
+      "loss": 0.8948,
+      "step": 591500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.990597651039464e-05,
+      "loss": 0.8786,
+      "step": 592000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9880559260825716e-05,
+      "loss": 0.884,
+      "step": 592500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985514201125679e-05,
+      "loss": 0.8809,
+      "step": 593000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982972476168787e-05,
+      "loss": 0.8821,
+      "step": 593500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9804307512118944e-05,
+      "loss": 0.872,
+      "step": 594000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9778890262550023e-05,
+      "loss": 0.8766,
+      "step": 594500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.97534730129811e-05,
+      "loss": 0.8832,
+      "step": 595000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9728055763412176e-05,
+      "loss": 0.8932,
+      "step": 595500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.970263851384325e-05,
+      "loss": 0.8759,
+      "step": 596000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.967722126427433e-05,
+      "loss": 0.8713,
+      "step": 596500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9651804014705405e-05,
+      "loss": 0.8616,
+      "step": 597000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.962638676513648e-05,
+      "loss": 0.8759,
+      "step": 597500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.960096951556756e-05,
+      "loss": 0.8907,
+      "step": 598000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9575552265998633e-05,
+      "loss": 0.8811,
+      "step": 598500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9550135016429712e-05,
+      "loss": 0.8743,
+      "step": 599000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9524717766860787e-05,
+      "loss": 0.8985,
+      "step": 599500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9499300517291862e-05,
+      "loss": 0.8669,
+      "step": 600000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.947388326772294e-05,
+      "loss": 0.8727,
+      "step": 600500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.944846601815402e-05,
+      "loss": 0.8729,
+      "step": 601000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9423048768585094e-05,
+      "loss": 0.8808,
+      "step": 601500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9397631519016172e-05,
+      "loss": 0.8698,
+      "step": 602000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9372214269447248e-05,
+      "loss": 0.8619,
+      "step": 602500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9346797019878323e-05,
+      "loss": 0.8647,
+      "step": 603000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.93213797703094e-05,
+      "loss": 0.8621,
+      "step": 603500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9295962520740476e-05,
+      "loss": 0.8481,
+      "step": 604000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.927054527117155e-05,
+      "loss": 0.8907,
+      "step": 604500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.924512802160263e-05,
+      "loss": 0.8663,
+      "step": 605000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9219710772033705e-05,
+      "loss": 0.8927,
+      "step": 605500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9194293522464783e-05,
+      "loss": 0.8841,
+      "step": 606000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.916887627289586e-05,
+      "loss": 0.8428,
+      "step": 606500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9143459023326933e-05,
+      "loss": 0.8604,
+      "step": 607000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9118041773758015e-05,
+      "loss": 0.868,
+      "step": 607500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.909262452418909e-05,
+      "loss": 0.8699,
+      "step": 608000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9067207274620165e-05,
+      "loss": 0.8722,
+      "step": 608500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9041790025051244e-05,
+      "loss": 0.898,
+      "step": 609000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.901637277548232e-05,
+      "loss": 0.8781,
+      "step": 609500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8990955525913394e-05,
+      "loss": 0.8858,
+      "step": 610000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8965538276344472e-05,
+      "loss": 0.8697,
+      "step": 610500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.8940121026775548e-05,
+      "loss": 0.8523,
+      "step": 611000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8914703777206623e-05,
+      "loss": 0.8754,
+      "step": 611500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.88892865276377e-05,
+      "loss": 0.8739,
+      "step": 612000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8863869278068776e-05,
+      "loss": 0.8813,
+      "step": 612500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8838452028499855e-05,
+      "loss": 0.881,
+      "step": 613000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.881303477893093e-05,
+      "loss": 0.8728,
+      "step": 613500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8787617529362008e-05,
+      "loss": 0.8992,
+      "step": 614000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8762200279793087e-05,
+      "loss": 0.8951,
+      "step": 614500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8736783030224162e-05,
+      "loss": 0.8731,
+      "step": 615000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8711365780655237e-05,
+      "loss": 0.883,
+      "step": 615500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8685948531086315e-05,
+      "loss": 0.8542,
+      "step": 616000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.866053128151739e-05,
+      "loss": 0.8832,
+      "step": 616500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8635114031948465e-05,
+      "loss": 0.8749,
+      "step": 617000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8609696782379544e-05,
+      "loss": 0.873,
+      "step": 617500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.858427953281062e-05,
+      "loss": 0.8564,
+      "step": 618000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8558862283241697e-05,
+      "loss": 0.8744,
+      "step": 618500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8533445033672773e-05,
+      "loss": 0.852,
+      "step": 619000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8508027784103848e-05,
+      "loss": 0.8596,
+      "step": 619500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8482610534534926e-05,
+      "loss": 0.8805,
+      "step": 620000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.8457193284966005e-05,
+      "loss": 0.8768,
+      "step": 620500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.843177603539708e-05,
+      "loss": 0.8765,
+      "step": 621000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8406358785828158e-05,
+      "loss": 0.8651,
+      "step": 621500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8380941536259233e-05,
+      "loss": 0.8575,
+      "step": 622000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8355524286690308e-05,
+      "loss": 0.8446,
+      "step": 622500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.8330107037121387e-05,
+      "loss": 0.867,
+      "step": 623000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8304689787552462e-05,
+      "loss": 0.8872,
+      "step": 623500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8279272537983537e-05,
+      "loss": 0.8591,
+      "step": 624000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8253855288414615e-05,
+      "loss": 0.8772,
+      "step": 624500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.822843803884569e-05,
+      "loss": 0.8681,
+      "step": 625000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.820302078927677e-05,
+      "loss": 0.8629,
+      "step": 625500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8177603539707844e-05,
+      "loss": 0.8682,
+      "step": 626000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.815218629013892e-05,
+      "loss": 0.8837,
+      "step": 626500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.812676904057e-05,
+      "loss": 0.8708,
+      "step": 627000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8101351791001076e-05,
+      "loss": 0.8496,
+      "step": 627500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.807593454143215e-05,
+      "loss": 0.8923,
+      "step": 628000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.805051729186323e-05,
+      "loss": 0.8705,
+      "step": 628500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8025100042294305e-05,
+      "loss": 0.8771,
+      "step": 629000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.799968279272538e-05,
+      "loss": 0.8706,
+      "step": 629500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7974265543156458e-05,
+      "loss": 0.8705,
+      "step": 630000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7948848293587533e-05,
+      "loss": 0.8488,
+      "step": 630500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7923431044018608e-05,
+      "loss": 0.8711,
+      "step": 631000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7898013794449687e-05,
+      "loss": 0.8783,
+      "step": 631500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7872596544880762e-05,
+      "loss": 0.8676,
+      "step": 632000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.784717929531184e-05,
+      "loss": 0.867,
+      "step": 632500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7821762045742915e-05,
+      "loss": 0.8812,
+      "step": 633000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7796344796173994e-05,
+      "loss": 0.8731,
+      "step": 633500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7770927546605072e-05,
+      "loss": 0.8679,
+      "step": 634000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7745510297036147e-05,
+      "loss": 0.8704,
+      "step": 634500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7720093047467222e-05,
+      "loss": 0.9008,
+      "step": 635000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.76946757978983e-05,
+      "loss": 0.8705,
+      "step": 635500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7669258548329376e-05,
+      "loss": 0.8719,
+      "step": 636000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.764384129876045e-05,
+      "loss": 0.8574,
+      "step": 636500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.761842404919153e-05,
+      "loss": 0.8807,
+      "step": 637000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7593006799622605e-05,
+      "loss": 0.8367,
+      "step": 637500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7567589550053683e-05,
+      "loss": 0.8661,
+      "step": 638000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7542172300484758e-05,
+      "loss": 0.8522,
+      "step": 638500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7516755050915833e-05,
+      "loss": 0.8639,
+      "step": 639000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.749133780134691e-05,
+      "loss": 0.8523,
+      "step": 639500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.746592055177799e-05,
+      "loss": 0.8616,
+      "step": 640000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7440503302209065e-05,
+      "loss": 0.8829,
+      "step": 640500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7415086052640144e-05,
+      "loss": 0.8585,
+      "step": 641000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.738966880307122e-05,
+      "loss": 0.8923,
+      "step": 641500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7364251553502294e-05,
+      "loss": 0.869,
+      "step": 642000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7338834303933372e-05,
+      "loss": 0.8842,
+      "step": 642500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.7313417054364447e-05,
+      "loss": 0.862,
+      "step": 643000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7287999804795522e-05,
+      "loss": 0.8681,
+      "step": 643500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.72625825552266e-05,
+      "loss": 0.8836,
+      "step": 644000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7237165305657676e-05,
+      "loss": 0.8728,
+      "step": 644500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7211748056088754e-05,
+      "loss": 0.8402,
+      "step": 645000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.718633080651983e-05,
+      "loss": 0.8719,
+      "step": 645500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7160913556950905e-05,
+      "loss": 0.8506,
+      "step": 646000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7135496307381983e-05,
+      "loss": 0.8545,
+      "step": 646500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.711007905781306e-05,
+      "loss": 0.8432,
+      "step": 647000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7084661808244137e-05,
+      "loss": 0.8623,
+      "step": 647500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7059244558675215e-05,
+      "loss": 0.8462,
+      "step": 648000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.703382730910629e-05,
+      "loss": 0.856,
+      "step": 648500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7008410059537365e-05,
+      "loss": 0.8672,
+      "step": 649000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6982992809968444e-05,
+      "loss": 0.8896,
+      "step": 649500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.695757556039952e-05,
+      "loss": 0.8698,
+      "step": 650000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6932158310830594e-05,
+      "loss": 0.8794,
+      "step": 650500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6906741061261672e-05,
+      "loss": 0.8828,
+      "step": 651000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6881323811692747e-05,
+      "loss": 0.843,
+      "step": 651500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6855906562123826e-05,
+      "loss": 0.8519,
+      "step": 652000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.68304893125549e-05,
+      "loss": 0.883,
+      "step": 652500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.680507206298598e-05,
+      "loss": 0.8715,
+      "step": 653000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6779654813417058e-05,
+      "loss": 0.8647,
+      "step": 653500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6754237563848133e-05,
+      "loss": 0.8638,
+      "step": 654000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6728820314279208e-05,
+      "loss": 0.8441,
+      "step": 654500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6703403064710286e-05,
+      "loss": 0.8532,
+      "step": 655000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.667798581514136e-05,
+      "loss": 0.8921,
+      "step": 655500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6652568565572437e-05,
+      "loss": 0.8621,
+      "step": 656000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6627151316003515e-05,
+      "loss": 0.8477,
+      "step": 656500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.660173406643459e-05,
+      "loss": 0.8837,
+      "step": 657000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6576316816865665e-05,
+      "loss": 0.8687,
+      "step": 657500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6550899567296744e-05,
+      "loss": 0.8526,
+      "step": 658000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.652548231772782e-05,
+      "loss": 0.8768,
+      "step": 658500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6500065068158897e-05,
+      "loss": 0.8684,
+      "step": 659000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6474647818589976e-05,
+      "loss": 0.8623,
+      "step": 659500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.644923056902105e-05,
+      "loss": 0.8685,
+      "step": 660000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.642381331945213e-05,
+      "loss": 0.8557,
+      "step": 660500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6398396069883204e-05,
+      "loss": 0.8522,
+      "step": 661000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.637297882031428e-05,
+      "loss": 0.8417,
+      "step": 661500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6347561570745358e-05,
+      "loss": 0.8581,
+      "step": 662000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6322144321176433e-05,
+      "loss": 0.8573,
+      "step": 662500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6296727071607508e-05,
+      "loss": 0.8415,
+      "step": 663000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6271309822038586e-05,
+      "loss": 0.8519,
+      "step": 663500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.624589257246966e-05,
+      "loss": 0.8821,
+      "step": 664000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.622047532290074e-05,
+      "loss": 0.8583,
+      "step": 664500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6195058073331815e-05,
+      "loss": 0.8534,
+      "step": 665000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.616964082376289e-05,
+      "loss": 0.874,
+      "step": 665500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.614422357419397e-05,
+      "loss": 0.8624,
+      "step": 666000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6118806324625047e-05,
+      "loss": 0.8739,
+      "step": 666500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6093389075056122e-05,
+      "loss": 0.8705,
+      "step": 667000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.60679718254872e-05,
+      "loss": 0.8661,
+      "step": 667500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6042554575918276e-05,
+      "loss": 0.8734,
+      "step": 668000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.601713732634935e-05,
+      "loss": 0.8659,
+      "step": 668500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.599172007678043e-05,
+      "loss": 0.8545,
+      "step": 669000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.5966302827211504e-05,
+      "loss": 0.8479,
+      "step": 669500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.594088557764258e-05,
+      "loss": 0.8695,
+      "step": 670000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5915468328073658e-05,
+      "loss": 0.867,
+      "step": 670500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5890051078504733e-05,
+      "loss": 0.8782,
+      "step": 671000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.586463382893581e-05,
+      "loss": 0.8418,
+      "step": 671500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5839216579366886e-05,
+      "loss": 0.8571,
+      "step": 672000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.5813799329797965e-05,
+      "loss": 0.8365,
+      "step": 672500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5788382080229043e-05,
+      "loss": 0.8393,
+      "step": 673000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.576296483066012e-05,
+      "loss": 0.8756,
+      "step": 673500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5737547581091193e-05,
+      "loss": 0.8588,
+      "step": 674000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.5712130331522272e-05,
+      "loss": 0.8568,
+      "step": 674500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5686713081953347e-05,
+      "loss": 0.8845,
+      "step": 675000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5661295832384422e-05,
+      "loss": 0.8774,
+      "step": 675500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.56358785828155e-05,
+      "loss": 0.8537,
+      "step": 676000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.5610461333246576e-05,
+      "loss": 0.8545,
+      "step": 676500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.558504408367765e-05,
+      "loss": 0.8647,
+      "step": 677000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.555962683410873e-05,
+      "loss": 0.8491,
+      "step": 677500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5534209584539804e-05,
+      "loss": 0.8465,
+      "step": 678000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5508792334970883e-05,
+      "loss": 0.8549,
+      "step": 678500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.548337508540196e-05,
+      "loss": 0.8629,
+      "step": 679000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.5457957835833036e-05,
+      "loss": 0.8899,
+      "step": 679500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5432540586264115e-05,
+      "loss": 0.8563,
+      "step": 680000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.540712333669519e-05,
+      "loss": 0.8613,
+      "step": 680500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5381706087126265e-05,
+      "loss": 0.8666,
+      "step": 681000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5356288837557343e-05,
+      "loss": 0.8715,
+      "step": 681500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.533087158798842e-05,
+      "loss": 0.8706,
+      "step": 682000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5305454338419493e-05,
+      "loss": 0.8498,
+      "step": 682500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5280037088850572e-05,
+      "loss": 0.863,
+      "step": 683000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5254619839281647e-05,
+      "loss": 0.8644,
+      "step": 683500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5229202589712724e-05,
+      "loss": 0.8593,
+      "step": 684000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.52037853401438e-05,
+      "loss": 0.8598,
+      "step": 684500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5178368090574879e-05,
+      "loss": 0.8518,
+      "step": 685000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5152950841005956e-05,
+      "loss": 0.8768,
+      "step": 685500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5127533591437033e-05,
+      "loss": 0.8381,
+      "step": 686000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5102116341868108e-05,
+      "loss": 0.8723,
+      "step": 686500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5076699092299184e-05,
+      "loss": 0.8697,
+      "step": 687000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5051281842730261e-05,
+      "loss": 0.854,
+      "step": 687500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5025864593161338e-05,
+      "loss": 0.8651,
+      "step": 688000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.5000447343592413e-05,
+      "loss": 0.857,
+      "step": 688500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.497503009402349e-05,
+      "loss": 0.8764,
+      "step": 689000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4949612844454567e-05,
+      "loss": 0.8667,
+      "step": 689500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4924195594885643e-05,
+      "loss": 0.8467,
+      "step": 690000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4898778345316718e-05,
+      "loss": 0.8752,
+      "step": 690500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4873361095747795e-05,
+      "loss": 0.863,
+      "step": 691000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.4847943846178874e-05,
+      "loss": 0.8538,
+      "step": 691500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.482252659660995e-05,
+      "loss": 0.8619,
+      "step": 692000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4797109347041027e-05,
+      "loss": 0.8544,
+      "step": 692500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4771692097472104e-05,
+      "loss": 0.8517,
+      "step": 693000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4746274847903179e-05,
+      "loss": 0.8614,
+      "step": 693500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4720857598334256e-05,
+      "loss": 0.8378,
+      "step": 694000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.4695440348765333e-05,
+      "loss": 0.8679,
+      "step": 694500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.467002309919641e-05,
+      "loss": 0.8594,
+      "step": 695000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4644605849627484e-05,
+      "loss": 0.8714,
+      "step": 695500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4619188600058561e-05,
+      "loss": 0.8709,
+      "step": 696000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4593771350489638e-05,
+      "loss": 0.8583,
+      "step": 696500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.4568354100920715e-05,
+      "loss": 0.8506,
+      "step": 697000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.454293685135179e-05,
+      "loss": 0.8481,
+      "step": 697500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.451751960178287e-05,
+      "loss": 0.8573,
+      "step": 698000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4492102352213947e-05,
+      "loss": 0.8384,
+      "step": 698500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4466685102645022e-05,
+      "loss": 0.8276,
+      "step": 699000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4441267853076099e-05,
+      "loss": 0.852,
+      "step": 699500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4415850603507175e-05,
+      "loss": 0.8628,
+      "step": 700000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4390433353938252e-05,
+      "loss": 0.8663,
+      "step": 700500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4365016104369327e-05,
+      "loss": 0.8569,
+      "step": 701000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4339598854800404e-05,
+      "loss": 0.8549,
+      "step": 701500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.431418160523148e-05,
+      "loss": 0.8469,
+      "step": 702000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4288764355662558e-05,
+      "loss": 0.8508,
+      "step": 702500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4263347106093633e-05,
+      "loss": 0.8339,
+      "step": 703000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.423792985652471e-05,
+      "loss": 0.8392,
+      "step": 703500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4212512606955786e-05,
+      "loss": 0.8725,
+      "step": 704000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4187095357386865e-05,
+      "loss": 0.8537,
+      "step": 704500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4161678107817941e-05,
+      "loss": 0.8463,
+      "step": 705000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4136260858249018e-05,
+      "loss": 0.8458,
+      "step": 705500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4110843608680093e-05,
+      "loss": 0.8314,
+      "step": 706000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.408542635911117e-05,
+      "loss": 0.855,
+      "step": 706500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4060009109542247e-05,
+      "loss": 0.8483,
+      "step": 707000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4034591859973324e-05,
+      "loss": 0.8695,
+      "step": 707500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4009174610404399e-05,
+      "loss": 0.8187,
+      "step": 708000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3983757360835475e-05,
+      "loss": 0.8329,
+      "step": 708500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3958340111266552e-05,
+      "loss": 0.8742,
+      "step": 709000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3932922861697629e-05,
+      "loss": 0.8799,
+      "step": 709500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3907505612128704e-05,
+      "loss": 0.8558,
+      "step": 710000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.388208836255978e-05,
+      "loss": 0.863,
+      "step": 710500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.385667111299086e-05,
+      "loss": 0.8657,
+      "step": 711000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3831253863421936e-05,
+      "loss": 0.8566,
+      "step": 711500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3805836613853013e-05,
+      "loss": 0.8568,
+      "step": 712000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.378041936428409e-05,
+      "loss": 0.8565,
+      "step": 712500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3755002114715165e-05,
+      "loss": 0.8495,
+      "step": 713000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3729584865146241e-05,
+      "loss": 0.8397,
+      "step": 713500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3704167615577318e-05,
+      "loss": 0.8439,
+      "step": 714000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.3678750366008395e-05,
+      "loss": 0.8332,
+      "step": 714500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.365333311643947e-05,
+      "loss": 0.8799,
+      "step": 715000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.3627915866870547e-05,
+      "loss": 0.8522,
+      "step": 715500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.3602498617301624e-05,
+      "loss": 0.8487,
+      "step": 716000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.35770813677327e-05,
+      "loss": 0.8685,
+      "step": 716500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3551664118163775e-05,
+      "loss": 0.8613,
+      "step": 717000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3526246868594856e-05,
+      "loss": 0.851,
+      "step": 717500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3500829619025932e-05,
+      "loss": 0.841,
+      "step": 718000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3475412369457007e-05,
+      "loss": 0.8565,
+      "step": 718500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3449995119888084e-05,
+      "loss": 0.8475,
+      "step": 719000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3424577870319161e-05,
+      "loss": 0.8371,
+      "step": 719500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3399160620750236e-05,
+      "loss": 0.8283,
+      "step": 720000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3373743371181313e-05,
+      "loss": 0.8539,
+      "step": 720500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.334832612161239e-05,
+      "loss": 0.8645,
+      "step": 721000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.3322908872043466e-05,
+      "loss": 0.8802,
+      "step": 721500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3297491622474541e-05,
+      "loss": 0.8409,
+      "step": 722000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3272074372905618e-05,
+      "loss": 0.8249,
+      "step": 722500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3246657123336695e-05,
+      "loss": 0.853,
+      "step": 723000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3221239873767772e-05,
+      "loss": 0.8368,
+      "step": 723500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.319582262419885e-05,
+      "loss": 0.8501,
+      "step": 724000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3170405374629927e-05,
+      "loss": 0.8321,
+      "step": 724500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3144988125061004e-05,
+      "loss": 0.8475,
+      "step": 725000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3119570875492079e-05,
+      "loss": 0.8664,
+      "step": 725500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3094153625923156e-05,
+      "loss": 0.875,
+      "step": 726000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3068736376354232e-05,
+      "loss": 0.8402,
+      "step": 726500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3043319126785309e-05,
+      "loss": 0.8383,
+      "step": 727000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3017901877216384e-05,
+      "loss": 0.8521,
+      "step": 727500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2992484627647461e-05,
+      "loss": 0.8454,
+      "step": 728000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2967067378078538e-05,
+      "loss": 0.8519,
+      "step": 728500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2941650128509614e-05,
+      "loss": 0.8681,
+      "step": 729000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.291623287894069e-05,
+      "loss": 0.8366,
+      "step": 729500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2890815629371766e-05,
+      "loss": 0.839,
+      "step": 730000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2865398379802845e-05,
+      "loss": 0.8379,
+      "step": 730500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2839981130233922e-05,
+      "loss": 0.8621,
+      "step": 731000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2814563880664998e-05,
+      "loss": 0.845,
+      "step": 731500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2789146631096075e-05,
+      "loss": 0.838,
+      "step": 732000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.276372938152715e-05,
+      "loss": 0.8704,
+      "step": 732500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2738312131958227e-05,
+      "loss": 0.8576,
+      "step": 733000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2712894882389304e-05,
+      "loss": 0.8638,
+      "step": 733500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.268747763282038e-05,
+      "loss": 0.8613,
+      "step": 734000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2662060383251456e-05,
+      "loss": 0.8667,
+      "step": 734500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2636643133682532e-05,
+      "loss": 0.851,
+      "step": 735000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2611225884113609e-05,
+      "loss": 0.8324,
+      "step": 735500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2585808634544686e-05,
+      "loss": 0.8722,
+      "step": 736000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2560391384975761e-05,
+      "loss": 0.8469,
+      "step": 736500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2534974135406841e-05,
+      "loss": 0.8328,
+      "step": 737000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2509556885837916e-05,
+      "loss": 0.8374,
+      "step": 737500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2484139636268991e-05,
+      "loss": 0.8629,
+      "step": 738000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2458722386700068e-05,
+      "loss": 0.8473,
+      "step": 738500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.2433305137131146e-05,
+      "loss": 0.8249,
+      "step": 739000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.2407887887562222e-05,
+      "loss": 0.8485,
+      "step": 739500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.2382470637993298e-05,
+      "loss": 0.8472,
+      "step": 740000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.2357053388424375e-05,
+      "loss": 0.8455,
+      "step": 740500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.2331636138855452e-05,
+      "loss": 0.8622,
+      "step": 741000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.2306218889286527e-05,
+      "loss": 0.85,
+      "step": 741500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.2280801639717605e-05,
+      "loss": 0.8409,
+      "step": 742000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.2255384390148682e-05,
+      "loss": 0.8779,
+      "step": 742500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.2229967140579757e-05,
+      "loss": 0.8424,
+      "step": 743000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.2204549891010834e-05,
+      "loss": 0.845,
+      "step": 743500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.217913264144191e-05,
+      "loss": 0.8418,
+      "step": 744000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.2153715391872988e-05,
+      "loss": 0.862,
+      "step": 744500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.2128298142304063e-05,
+      "loss": 0.8366,
+      "step": 745000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.2102880892735141e-05,
+      "loss": 0.8596,
+      "step": 745500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.2077463643166218e-05,
+      "loss": 0.8502,
+      "step": 746000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.2052046393597295e-05,
+      "loss": 0.8534,
+      "step": 746500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.202662914402837e-05,
+      "loss": 0.8485,
+      "step": 747000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.2001211894459446e-05,
+      "loss": 0.8249,
+      "step": 747500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.1975794644890523e-05,
+      "loss": 0.8498,
+      "step": 748000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.19503773953216e-05,
+      "loss": 0.8419,
+      "step": 748500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.1924960145752677e-05,
+      "loss": 0.8535,
+      "step": 749000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.1899542896183754e-05,
+      "loss": 0.8394,
+      "step": 749500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.187412564661483e-05,
+      "loss": 0.8352,
+      "step": 750000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.1848708397045905e-05,
+      "loss": 0.8314,
+      "step": 750500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.1823291147476982e-05,
+      "loss": 0.8549,
+      "step": 751000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.1797873897908059e-05,
+      "loss": 0.8445,
+      "step": 751500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.1772456648339136e-05,
+      "loss": 0.8339,
+      "step": 752000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.1747039398770212e-05,
+      "loss": 0.8329,
+      "step": 752500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.172162214920129e-05,
+      "loss": 0.8328,
+      "step": 753000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.1696204899632366e-05,
+      "loss": 0.8628,
+      "step": 753500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.1670787650063441e-05,
+      "loss": 0.829,
+      "step": 754000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.1645370400494518e-05,
+      "loss": 0.8302,
+      "step": 754500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.1619953150925596e-05,
+      "loss": 0.8335,
+      "step": 755000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.1594535901356671e-05,
+      "loss": 0.8532,
+      "step": 755500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.1569118651787748e-05,
+      "loss": 0.846,
+      "step": 756000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.1543701402218825e-05,
+      "loss": 0.8421,
+      "step": 756500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.1518284152649902e-05,
+      "loss": 0.852,
+      "step": 757000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.1492866903080977e-05,
+      "loss": 0.8502,
+      "step": 757500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.1467449653512054e-05,
+      "loss": 0.8351,
+      "step": 758000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.1442032403943132e-05,
+      "loss": 0.8397,
+      "step": 758500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.1416615154374207e-05,
+      "loss": 0.8392,
+      "step": 759000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.1391197904805284e-05,
+      "loss": 0.8367,
+      "step": 759500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.136578065523636e-05,
+      "loss": 0.843,
+      "step": 760000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.1340363405667437e-05,
+      "loss": 0.8487,
+      "step": 760500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.1314946156098512e-05,
+      "loss": 0.8635,
+      "step": 761000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.1289528906529591e-05,
+      "loss": 0.8464,
+      "step": 761500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.1264111656960668e-05,
+      "loss": 0.8616,
+      "step": 762000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.1238694407391743e-05,
+      "loss": 0.8266,
+      "step": 762500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.121327715782282e-05,
+      "loss": 0.8357,
+      "step": 763000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.1187859908253896e-05,
+      "loss": 0.8362,
+      "step": 763500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1162442658684973e-05,
+      "loss": 0.8298,
+      "step": 764000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1137025409116048e-05,
+      "loss": 0.8595,
+      "step": 764500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1111608159547127e-05,
+      "loss": 0.8624,
+      "step": 765000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.1086190909978203e-05,
+      "loss": 0.8169,
+      "step": 765500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.1060773660409278e-05,
+      "loss": 0.8431,
+      "step": 766000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.1035356410840355e-05,
+      "loss": 0.8523,
+      "step": 766500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.1009939161271432e-05,
+      "loss": 0.8312,
+      "step": 767000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.0984521911702509e-05,
+      "loss": 0.8378,
+      "step": 767500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.0959104662133586e-05,
+      "loss": 0.8605,
+      "step": 768000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.0933687412564662e-05,
+      "loss": 0.8342,
+      "step": 768500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.0908270162995739e-05,
+      "loss": 0.8231,
+      "step": 769000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.0882852913426816e-05,
+      "loss": 0.8605,
+      "step": 769500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.0857435663857891e-05,
+      "loss": 0.8333,
+      "step": 770000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.0832018414288968e-05,
+      "loss": 0.846,
+      "step": 770500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.0806601164720044e-05,
+      "loss": 0.8449,
+      "step": 771000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.0781183915151121e-05,
+      "loss": 0.8495,
+      "step": 771500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.0755766665582198e-05,
+      "loss": 0.8575,
+      "step": 772000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.0730349416013275e-05,
+      "loss": 0.8204,
+      "step": 772500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.0704932166444352e-05,
+      "loss": 0.8368,
+      "step": 773000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0679514916875427e-05,
+      "loss": 0.8246,
+      "step": 773500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0654097667306503e-05,
+      "loss": 0.8263,
+      "step": 774000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0628680417737582e-05,
+      "loss": 0.8463,
+      "step": 774500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0603263168168657e-05,
+      "loss": 0.834,
+      "step": 775000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.0577845918599734e-05,
+      "loss": 0.8306,
+      "step": 775500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.055242866903081e-05,
+      "loss": 0.8362,
+      "step": 776000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0527011419461887e-05,
+      "loss": 0.8444,
+      "step": 776500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0501594169892962e-05,
+      "loss": 0.8307,
+      "step": 777000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0476176920324039e-05,
+      "loss": 0.8121,
+      "step": 777500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.0450759670755118e-05,
+      "loss": 0.8404,
+      "step": 778000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0425342421186193e-05,
+      "loss": 0.8547,
+      "step": 778500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.039992517161727e-05,
+      "loss": 0.8437,
+      "step": 779000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0374507922048346e-05,
+      "loss": 0.8385,
+      "step": 779500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0349090672479423e-05,
+      "loss": 0.8425,
+      "step": 780000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.0323673422910498e-05,
+      "loss": 0.8349,
+      "step": 780500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0298256173341577e-05,
+      "loss": 0.8482,
+      "step": 781000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0272838923772653e-05,
+      "loss": 0.8378,
+      "step": 781500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0247421674203728e-05,
+      "loss": 0.8382,
+      "step": 782000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0222004424634805e-05,
+      "loss": 0.8386,
+      "step": 782500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.0196587175065882e-05,
+      "loss": 0.8435,
+      "step": 783000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0171169925496959e-05,
+      "loss": 0.8362,
+      "step": 783500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0145752675928034e-05,
+      "loss": 0.8264,
+      "step": 784000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0120335426359112e-05,
+      "loss": 0.8325,
+      "step": 784500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0094918176790189e-05,
+      "loss": 0.8482,
+      "step": 785000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.0069500927221264e-05,
+      "loss": 0.8466,
+      "step": 785500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.004408367765234e-05,
+      "loss": 0.8221,
+      "step": 786000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.0018666428083418e-05,
+      "loss": 0.8484,
+      "step": 786500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.993249178514494e-06,
+      "loss": 0.8448,
+      "step": 787000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.967831928945571e-06,
+      "loss": 0.8297,
+      "step": 787500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.942414679376648e-06,
+      "loss": 0.8479,
+      "step": 788000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.916997429807725e-06,
+      "loss": 0.8537,
+      "step": 788500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.8915801802388e-06,
+      "loss": 0.8216,
+      "step": 789000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.866162930669877e-06,
+      "loss": 0.8362,
+      "step": 789500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.840745681100953e-06,
+      "loss": 0.823,
+      "step": 790000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.81532843153203e-06,
+      "loss": 0.8302,
+      "step": 790500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.789911181963107e-06,
+      "loss": 0.8344,
+      "step": 791000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.764493932394184e-06,
+      "loss": 0.8635,
+      "step": 791500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.73907668282526e-06,
+      "loss": 0.82,
+      "step": 792000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.713659433256337e-06,
+      "loss": 0.8594,
+      "step": 792500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.688242183687412e-06,
+      "loss": 0.8303,
+      "step": 793000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.662824934118489e-06,
+      "loss": 0.829,
+      "step": 793500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.637407684549567e-06,
+      "loss": 0.8398,
+      "step": 794000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.611990434980643e-06,
+      "loss": 0.8208,
+      "step": 794500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.58657318541172e-06,
+      "loss": 0.8291,
+      "step": 795000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.561155935842796e-06,
+      "loss": 0.8321,
+      "step": 795500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.535738686273873e-06,
+      "loss": 0.8369,
+      "step": 796000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.510321436704948e-06,
+      "loss": 0.8214,
+      "step": 796500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.484904187136026e-06,
+      "loss": 0.8149,
+      "step": 797000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.459486937567103e-06,
+      "loss": 0.8332,
+      "step": 797500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.434069687998178e-06,
+      "loss": 0.8433,
+      "step": 798000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.408652438429255e-06,
+      "loss": 0.8303,
+      "step": 798500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.383235188860332e-06,
+      "loss": 0.8063,
+      "step": 799000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.357817939291409e-06,
+      "loss": 0.8428,
+      "step": 799500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.332400689722484e-06,
+      "loss": 0.8211,
+      "step": 800000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.306983440153562e-06,
+      "loss": 0.8224,
+      "step": 800500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.281566190584639e-06,
+      "loss": 0.8408,
+      "step": 801000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.256148941015714e-06,
+      "loss": 0.815,
+      "step": 801500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.23073169144679e-06,
+      "loss": 0.8395,
+      "step": 802000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.205314441877867e-06,
+      "loss": 0.8267,
+      "step": 802500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.179897192308944e-06,
+      "loss": 0.8677,
+      "step": 803000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.154479942740021e-06,
+      "loss": 0.8489,
+      "step": 803500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.129062693171098e-06,
+      "loss": 0.8317,
+      "step": 804000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.103645443602175e-06,
+      "loss": 0.8482,
+      "step": 804500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.07822819403325e-06,
+      "loss": 0.8245,
+      "step": 805000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.052810944464326e-06,
+      "loss": 0.8309,
+      "step": 805500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.027393694895403e-06,
+      "loss": 0.8591,
+      "step": 806000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.00197644532648e-06,
+      "loss": 0.8305,
+      "step": 806500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.976559195757557e-06,
+      "loss": 0.8414,
+      "step": 807000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.951141946188633e-06,
+      "loss": 0.8415,
+      "step": 807500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.92572469661971e-06,
+      "loss": 0.8436,
+      "step": 808000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.900307447050785e-06,
+      "loss": 0.8516,
+      "step": 808500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.874890197481862e-06,
+      "loss": 0.837,
+      "step": 809000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.849472947912939e-06,
+      "loss": 0.8496,
+      "step": 809500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.824055698344017e-06,
+      "loss": 0.8428,
+      "step": 810000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.798638448775092e-06,
+      "loss": 0.8455,
+      "step": 810500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.773221199206169e-06,
+      "loss": 0.8282,
+      "step": 811000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.747803949637246e-06,
+      "loss": 0.8335,
+      "step": 811500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.722386700068321e-06,
+      "loss": 0.818,
+      "step": 812000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.696969450499398e-06,
+      "loss": 0.8299,
+      "step": 812500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.671552200930475e-06,
+      "loss": 0.8182,
+      "step": 813000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.646134951361553e-06,
+      "loss": 0.8507,
+      "step": 813500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.620717701792628e-06,
+      "loss": 0.8342,
+      "step": 814000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.595300452223705e-06,
+      "loss": 0.8244,
+      "step": 814500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.569883202654782e-06,
+      "loss": 0.8231,
+      "step": 815000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.544465953085858e-06,
+      "loss": 0.8236,
+      "step": 815500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.519048703516933e-06,
+      "loss": 0.8365,
+      "step": 816000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.493631453948012e-06,
+      "loss": 0.8345,
+      "step": 816500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.468214204379089e-06,
+      "loss": 0.8258,
+      "step": 817000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.442796954810164e-06,
+      "loss": 0.8365,
+      "step": 817500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.41737970524124e-06,
+      "loss": 0.8312,
+      "step": 818000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.391962455672317e-06,
+      "loss": 0.828,
+      "step": 818500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.366545206103394e-06,
+      "loss": 0.8307,
+      "step": 819000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.341127956534469e-06,
+      "loss": 0.8613,
+      "step": 819500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.315710706965548e-06,
+      "loss": 0.8539,
+      "step": 820000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.290293457396624e-06,
+      "loss": 0.8288,
+      "step": 820500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.2648762078277e-06,
+      "loss": 0.8498,
+      "step": 821000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.239458958258776e-06,
+      "loss": 0.8433,
+      "step": 821500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.214041708689853e-06,
+      "loss": 0.8434,
+      "step": 822000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.18862445912093e-06,
+      "loss": 0.8328,
+      "step": 822500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.163207209552007e-06,
+      "loss": 0.8382,
+      "step": 823000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.137789959983083e-06,
+      "loss": 0.8438,
+      "step": 823500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.11237271041416e-06,
+      "loss": 0.826,
+      "step": 824000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.086955460845235e-06,
+      "loss": 0.8354,
+      "step": 824500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.061538211276312e-06,
+      "loss": 0.8237,
+      "step": 825000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.036120961707389e-06,
+      "loss": 0.8373,
+      "step": 825500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.010703712138465e-06,
+      "loss": 0.8285,
+      "step": 826000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.985286462569542e-06,
+      "loss": 0.8583,
+      "step": 826500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.959869213000619e-06,
+      "loss": 0.8424,
+      "step": 827000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.934451963431696e-06,
+      "loss": 0.8357,
+      "step": 827500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.909034713862771e-06,
+      "loss": 0.8296,
+      "step": 828000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.883617464293848e-06,
+      "loss": 0.8138,
+      "step": 828500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.858200214724924e-06,
+      "loss": 0.8288,
+      "step": 829000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.832782965156001e-06,
+      "loss": 0.8418,
+      "step": 829500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.807365715587078e-06,
+      "loss": 0.8213,
+      "step": 830000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.781948466018155e-06,
+      "loss": 0.8087,
+      "step": 830500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.756531216449231e-06,
+      "loss": 0.8187,
+      "step": 831000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.731113966880307e-06,
+      "loss": 0.8336,
+      "step": 831500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.705696717311383e-06,
+      "loss": 0.8215,
+      "step": 832000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.68027946774246e-06,
+      "loss": 0.8109,
+      "step": 832500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.654862218173539e-06,
+      "loss": 0.8182,
+      "step": 833000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.629444968604614e-06,
+      "loss": 0.8478,
+      "step": 833500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.60402771903569e-06,
+      "loss": 0.8223,
+      "step": 834000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.578610469466766e-06,
+      "loss": 0.8186,
+      "step": 834500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.553193219897843e-06,
+      "loss": 0.8214,
+      "step": 835000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.527775970328919e-06,
+      "loss": 0.8418,
+      "step": 835500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.5023587207599975e-06,
+      "loss": 0.8363,
+      "step": 836000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.476941471191073e-06,
+      "loss": 0.8204,
+      "step": 836500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.45152422162215e-06,
+      "loss": 0.8076,
+      "step": 837000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.426106972053226e-06,
+      "loss": 0.8186,
+      "step": 837500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.400689722484303e-06,
+      "loss": 0.8264,
+      "step": 838000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.375272472915379e-06,
+      "loss": 0.8314,
+      "step": 838500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.3498552233464556e-06,
+      "loss": 0.8159,
+      "step": 839000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.324437973777533e-06,
+      "loss": 0.8461,
+      "step": 839500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.299020724208609e-06,
+      "loss": 0.8142,
+      "step": 840000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.273603474639686e-06,
+      "loss": 0.8275,
+      "step": 840500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.248186225070762e-06,
+      "loss": 0.8527,
+      "step": 841000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.2227689755018386e-06,
+      "loss": 0.8077,
+      "step": 841500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.1973517259329145e-06,
+      "loss": 0.8188,
+      "step": 842000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.171934476363992e-06,
+      "loss": 0.8299,
+      "step": 842500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.146517226795069e-06,
+      "loss": 0.8478,
+      "step": 843000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.121099977226145e-06,
+      "loss": 0.8292,
+      "step": 843500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.0956827276572216e-06,
+      "loss": 0.8189,
+      "step": 844000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.0702654780882975e-06,
+      "loss": 0.8111,
+      "step": 844500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.044848228519374e-06,
+      "loss": 0.8468,
+      "step": 845000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.01943097895045e-06,
+      "loss": 0.8247,
+      "step": 845500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.994013729381528e-06,
+      "loss": 0.8546,
+      "step": 846000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.9685964798126046e-06,
+      "loss": 0.8307,
+      "step": 846500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.9431792302436805e-06,
+      "loss": 0.8178,
+      "step": 847000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.917761980674757e-06,
+      "loss": 0.838,
+      "step": 847500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.892344731105833e-06,
+      "loss": 0.8243,
+      "step": 848000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.86692748153691e-06,
+      "loss": 0.8384,
+      "step": 848500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.8415102319679876e-06,
+      "loss": 0.8214,
+      "step": 849000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.8160929823990635e-06,
+      "loss": 0.8275,
+      "step": 849500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.79067573283014e-06,
+      "loss": 0.8174,
+      "step": 850000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.765258483261216e-06,
+      "loss": 0.8135,
+      "step": 850500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.739841233692293e-06,
+      "loss": 0.8048,
+      "step": 851000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.714423984123369e-06,
+      "loss": 0.8255,
+      "step": 851500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.689006734554446e-06,
+      "loss": 0.8431,
+      "step": 852000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.663589484985523e-06,
+      "loss": 0.8463,
+      "step": 852500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.638172235416599e-06,
+      "loss": 0.843,
+      "step": 853000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.612754985847676e-06,
+      "loss": 0.8059,
+      "step": 853500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.587337736278752e-06,
+      "loss": 0.8372,
+      "step": 854000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.561920486709829e-06,
+      "loss": 0.8298,
+      "step": 854500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.5365032371409046e-06,
+      "loss": 0.8103,
+      "step": 855000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.511085987571983e-06,
+      "loss": 0.8168,
+      "step": 855500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.485668738003059e-06,
+      "loss": 0.8391,
+      "step": 856000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.460251488434135e-06,
+      "loss": 0.8212,
+      "step": 856500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.434834238865212e-06,
+      "loss": 0.8105,
+      "step": 857000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.4094169892962876e-06,
+      "loss": 0.8338,
+      "step": 857500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.383999739727364e-06,
+      "loss": 0.8277,
+      "step": 858000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.35858249015844e-06,
+      "loss": 0.8206,
+      "step": 858500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.333165240589519e-06,
+      "loss": 0.8328,
+      "step": 859000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.307747991020595e-06,
+      "loss": 0.8152,
+      "step": 859500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.282330741451671e-06,
+      "loss": 0.8139,
+      "step": 860000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.256913491882747e-06,
+      "loss": 0.8311,
+      "step": 860500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.231496242313824e-06,
+      "loss": 0.8095,
+      "step": 861000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.206078992744901e-06,
+      "loss": 0.8359,
+      "step": 861500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.180661743175977e-06,
+      "loss": 0.8203,
+      "step": 862000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.1552444936070536e-06,
+      "loss": 0.84,
+      "step": 862500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.12982724403813e-06,
+      "loss": 0.8124,
+      "step": 863000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.104409994469207e-06,
+      "loss": 0.8331,
+      "step": 863500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.078992744900283e-06,
+      "loss": 0.8334,
+      "step": 864000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.05357549533136e-06,
+      "loss": 0.8461,
+      "step": 864500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.028158245762437e-06,
+      "loss": 0.8518,
+      "step": 865000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.0027409961935125e-06,
+      "loss": 0.8265,
+      "step": 865500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.97732374662459e-06,
+      "loss": 0.8121,
+      "step": 866000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.951906497055666e-06,
+      "loss": 0.8326,
+      "step": 866500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.926489247486743e-06,
+      "loss": 0.8336,
+      "step": 867000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.901071997917819e-06,
+      "loss": 0.8327,
+      "step": 867500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8756547483488955e-06,
+      "loss": 0.8019,
+      "step": 868000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.850237498779972e-06,
+      "loss": 0.8042,
+      "step": 868500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.824820249211048e-06,
+      "loss": 0.8482,
+      "step": 869000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.799402999642126e-06,
+      "loss": 0.8178,
+      "step": 869500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.773985750073202e-06,
+      "loss": 0.8266,
+      "step": 870000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.7485685005042785e-06,
+      "loss": 0.8353,
+      "step": 870500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.723151250935355e-06,
+      "loss": 0.838,
+      "step": 871000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.697734001366432e-06,
+      "loss": 0.8174,
+      "step": 871500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.672316751797508e-06,
+      "loss": 0.8322,
+      "step": 872000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.646899502228585e-06,
+      "loss": 0.8121,
+      "step": 872500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.6214822526596615e-06,
+      "loss": 0.8408,
+      "step": 873000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5960650030907374e-06,
+      "loss": 0.8145,
+      "step": 873500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.570647753521814e-06,
+      "loss": 0.8074,
+      "step": 874000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.545230503952891e-06,
+      "loss": 0.8098,
+      "step": 874500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.519813254383968e-06,
+      "loss": 0.8243,
+      "step": 875000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.494396004815044e-06,
+      "loss": 0.8411,
+      "step": 875500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.4689787552461204e-06,
+      "loss": 0.8303,
+      "step": 876000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.443561505677197e-06,
+      "loss": 0.8298,
+      "step": 876500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.418144256108273e-06,
+      "loss": 0.8218,
+      "step": 877000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.392727006539351e-06,
+      "loss": 0.8098,
+      "step": 877500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.367309756970427e-06,
+      "loss": 0.8238,
+      "step": 878000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.3418925074015034e-06,
+      "loss": 0.8,
+      "step": 878500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.31647525783258e-06,
+      "loss": 0.8223,
+      "step": 879000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.291058008263656e-06,
+      "loss": 0.8346,
+      "step": 879500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.265640758694733e-06,
+      "loss": 0.8084,
+      "step": 880000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.240223509125809e-06,
+      "loss": 0.8227,
+      "step": 880500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.2148062595568864e-06,
+      "loss": 0.8239,
+      "step": 881000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.189389009987962e-06,
+      "loss": 0.8204,
+      "step": 881500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.163971760419039e-06,
+      "loss": 0.8275,
+      "step": 882000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.138554510850116e-06,
+      "loss": 0.8432,
+      "step": 882500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.113137261281193e-06,
+      "loss": 0.8237,
+      "step": 883000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.087720011712269e-06,
+      "loss": 0.8096,
+      "step": 883500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.062302762143345e-06,
+      "loss": 0.8104,
+      "step": 884000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.036885512574422e-06,
+      "loss": 0.8155,
+      "step": 884500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.011468263005498e-06,
+      "loss": 0.8371,
+      "step": 885000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.986051013436576e-06,
+      "loss": 0.8315,
+      "step": 885500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.960633763867652e-06,
+      "loss": 0.7936,
+      "step": 886000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.935216514298728e-06,
+      "loss": 0.8248,
+      "step": 886500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.909799264729804e-06,
+      "loss": 0.8245,
+      "step": 887000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.884382015160881e-06,
+      "loss": 0.8116,
+      "step": 887500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.858964765591958e-06,
+      "loss": 0.8181,
+      "step": 888000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.833547516023034e-06,
+      "loss": 0.8577,
+      "step": 888500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.808130266454111e-06,
+      "loss": 0.8207,
+      "step": 889000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.782713016885187e-06,
+      "loss": 0.7982,
+      "step": 889500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.757295767316264e-06,
+      "loss": 0.7982,
+      "step": 890000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.731878517747341e-06,
+      "loss": 0.8099,
+      "step": 890500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.706461268178417e-06,
+      "loss": 0.832,
+      "step": 891000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.6810440186094935e-06,
+      "loss": 0.8152,
+      "step": 891500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.65562676904057e-06,
+      "loss": 0.8283,
+      "step": 892000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.630209519471647e-06,
+      "loss": 0.8213,
+      "step": 892500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.604792269902723e-06,
+      "loss": 0.8185,
+      "step": 893000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5793750203338e-06,
+      "loss": 0.8455,
+      "step": 893500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.5539577707648765e-06,
+      "loss": 0.808,
+      "step": 894000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.528540521195953e-06,
+      "loss": 0.8318,
+      "step": 894500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.503123271627029e-06,
+      "loss": 0.8324,
+      "step": 895000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.477706022058106e-06,
+      "loss": 0.8288,
+      "step": 895500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.452288772489183e-06,
+      "loss": 0.8068,
+      "step": 896000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.426871522920259e-06,
+      "loss": 0.8125,
+      "step": 896500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.401454273351336e-06,
+      "loss": 0.8088,
+      "step": 897000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.376037023782412e-06,
+      "loss": 0.8249,
+      "step": 897500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.350619774213489e-06,
+      "loss": 0.8066,
+      "step": 898000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.325202524644566e-06,
+      "loss": 0.8373,
+      "step": 898500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.299785275075642e-06,
+      "loss": 0.8083,
+      "step": 899000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.2743680255067185e-06,
+      "loss": 0.829,
+      "step": 899500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.248950775937794e-06,
+      "loss": 0.7986,
+      "step": 900000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.223533526368872e-06,
+      "loss": 0.8297,
+      "step": 900500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.198116276799948e-06,
+      "loss": 0.8396,
+      "step": 901000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.172699027231025e-06,
+      "loss": 0.8081,
+      "step": 901500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.1472817776621015e-06,
+      "loss": 0.8464,
+      "step": 902000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.121864528093177e-06,
+      "loss": 0.7914,
+      "step": 902500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.096447278524254e-06,
+      "loss": 0.8123,
+      "step": 903000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.071030028955331e-06,
+      "loss": 0.8476,
+      "step": 903500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.045612779386408e-06,
+      "loss": 0.8375,
+      "step": 904000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.020195529817484e-06,
+      "loss": 0.8075,
+      "step": 904500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.994778280248561e-06,
+      "loss": 0.8422,
+      "step": 905000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.969361030679637e-06,
+      "loss": 0.8156,
+      "step": 905500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.943943781110714e-06,
+      "loss": 0.8235,
+      "step": 906000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.91852653154179e-06,
+      "loss": 0.8428,
+      "step": 906500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.893109281972867e-06,
+      "loss": 0.8243,
+      "step": 907000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.867692032403943e-06,
+      "loss": 0.8262,
+      "step": 907500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.842274782835019e-06,
+      "loss": 0.8435,
+      "step": 908000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.816857533266097e-06,
+      "loss": 0.8359,
+      "step": 908500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.791440283697173e-06,
+      "loss": 0.8216,
+      "step": 909000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.766023034128249e-06,
+      "loss": 0.8286,
+      "step": 909500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7406057845593264e-06,
+      "loss": 0.8098,
+      "step": 910000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7151885349904027e-06,
+      "loss": 0.8099,
+      "step": 910500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.689771285421479e-06,
+      "loss": 0.8256,
+      "step": 911000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.664354035852556e-06,
+      "loss": 0.8189,
+      "step": 911500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.638936786283632e-06,
+      "loss": 0.8135,
+      "step": 912000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.6135195367147085e-06,
+      "loss": 0.8293,
+      "step": 912500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.588102287145785e-06,
+      "loss": 0.8122,
+      "step": 913000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.562685037576862e-06,
+      "loss": 0.8176,
+      "step": 913500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.5372677880079384e-06,
+      "loss": 0.8138,
+      "step": 914000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.5118505384390148e-06,
+      "loss": 0.8392,
+      "step": 914500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.486433288870092e-06,
+      "loss": 0.8362,
+      "step": 915000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.4610160393011683e-06,
+      "loss": 0.8134,
+      "step": 915500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.4355987897322447e-06,
+      "loss": 0.8206,
+      "step": 916000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.4101815401633214e-06,
+      "loss": 0.8163,
+      "step": 916500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.3847642905943978e-06,
+      "loss": 0.8204,
+      "step": 917000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.359347041025474e-06,
+      "loss": 0.826,
+      "step": 917500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.3339297914565513e-06,
+      "loss": 0.8128,
+      "step": 918000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.3085125418876277e-06,
+      "loss": 0.8353,
+      "step": 918500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.283095292318704e-06,
+      "loss": 0.8294,
+      "step": 919000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.2576780427497804e-06,
+      "loss": 0.7966,
+      "step": 919500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.232260793180857e-06,
+      "loss": 0.8098,
+      "step": 920000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.2068435436119335e-06,
+      "loss": 0.8475,
+      "step": 920500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.18142629404301e-06,
+      "loss": 0.8088,
+      "step": 921000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.156009044474087e-06,
+      "loss": 0.8287,
+      "step": 921500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1305917949051634e-06,
+      "loss": 0.8267,
+      "step": 922000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.10517454533624e-06,
+      "loss": 0.8327,
+      "step": 922500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0797572957673165e-06,
+      "loss": 0.809,
+      "step": 923000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054340046198393e-06,
+      "loss": 0.8156,
+      "step": 923500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.028922796629469e-06,
+      "loss": 0.8068,
+      "step": 924000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.003505547060546e-06,
+      "loss": 0.8308,
+      "step": 924500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.9780882974916227e-06,
+      "loss": 0.8099,
+      "step": 925000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.952671047922699e-06,
+      "loss": 0.8105,
+      "step": 925500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.927253798353776e-06,
+      "loss": 0.848,
+      "step": 926000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9018365487848526e-06,
+      "loss": 0.8339,
+      "step": 926500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.876419299215929e-06,
+      "loss": 0.8207,
+      "step": 927000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8510020496470053e-06,
+      "loss": 0.841,
+      "step": 927500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8255848000780816e-06,
+      "loss": 0.8053,
+      "step": 928000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8001675505091584e-06,
+      "loss": 0.8186,
+      "step": 928500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.774750300940235e-06,
+      "loss": 0.7948,
+      "step": 929000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.7493330513713115e-06,
+      "loss": 0.8268,
+      "step": 929500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.7239158018023883e-06,
+      "loss": 0.8221,
+      "step": 930000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.6984985522334646e-06,
+      "loss": 0.839,
+      "step": 930500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.673081302664541e-06,
+      "loss": 0.8139,
+      "step": 931000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.6476640530956178e-06,
+      "loss": 0.7891,
+      "step": 931500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.622246803526694e-06,
+      "loss": 0.8337,
+      "step": 932000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.596829553957771e-06,
+      "loss": 0.8222,
+      "step": 932500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.5714123043888476e-06,
+      "loss": 0.8257,
+      "step": 933000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.545995054819924e-06,
+      "loss": 0.843,
+      "step": 933500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.5205778052510008e-06,
+      "loss": 0.8166,
+      "step": 934000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.495160555682077e-06,
+      "loss": 0.8155,
+      "step": 934500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.4697433061131535e-06,
+      "loss": 0.817,
+      "step": 935000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.4443260565442302e-06,
+      "loss": 0.8164,
+      "step": 935500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.4189088069753066e-06,
+      "loss": 0.8301,
+      "step": 936000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.3934915574063833e-06,
+      "loss": 0.8336,
+      "step": 936500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.36807430783746e-06,
+      "loss": 0.802,
+      "step": 937000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.3426570582685365e-06,
+      "loss": 0.8027,
+      "step": 937500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.3172398086996132e-06,
+      "loss": 0.7946,
+      "step": 938000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.2918225591306896e-06,
+      "loss": 0.8373,
+      "step": 938500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.266405309561766e-06,
+      "loss": 0.8374,
+      "step": 939000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.2409880599928427e-06,
+      "loss": 0.8279,
+      "step": 939500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.215570810423919e-06,
+      "loss": 0.8112,
+      "step": 940000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.190153560854996e-06,
+      "loss": 0.8044,
+      "step": 940500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.164736311286072e-06,
+      "loss": 0.7957,
+      "step": 941000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.139319061717149e-06,
+      "loss": 0.8294,
+      "step": 941500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.1139018121482257e-06,
+      "loss": 0.8289,
+      "step": 942000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.0884845625793016e-06,
+      "loss": 0.8225,
+      "step": 942500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.0630673130103784e-06,
+      "loss": 0.8064,
+      "step": 943000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.037650063441455e-06,
+      "loss": 0.8231,
+      "step": 943500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.0122328138725315e-06,
+      "loss": 0.8142,
+      "step": 944000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.9868155643036083e-06,
+      "loss": 0.8085,
+      "step": 944500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.9613983147346846e-06,
+      "loss": 0.8251,
+      "step": 945000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.9359810651657614e-06,
+      "loss": 0.8123,
+      "step": 945500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.9105638155968377e-06,
+      "loss": 0.8362,
+      "step": 946000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8851465660279143e-06,
+      "loss": 0.8129,
+      "step": 946500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8597293164589908e-06,
+      "loss": 0.8213,
+      "step": 947000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.8343120668900672e-06,
+      "loss": 0.8065,
+      "step": 947500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.808894817321144e-06,
+      "loss": 0.8152,
+      "step": 948000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7834775677522207e-06,
+      "loss": 0.8148,
+      "step": 948500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.758060318183297e-06,
+      "loss": 0.8132,
+      "step": 949000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7326430686143736e-06,
+      "loss": 0.8409,
+      "step": 949500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7072258190454504e-06,
+      "loss": 0.8101,
+      "step": 950000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.6818085694765268e-06,
+      "loss": 0.8084,
+      "step": 950500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.6563913199076033e-06,
+      "loss": 0.8196,
+      "step": 951000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.6309740703386797e-06,
+      "loss": 0.82,
+      "step": 951500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.6055568207697564e-06,
+      "loss": 0.7967,
+      "step": 952000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.580139571200833e-06,
+      "loss": 0.8143,
+      "step": 952500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5547223216319093e-06,
+      "loss": 0.8138,
+      "step": 953000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.529305072062986e-06,
+      "loss": 0.8168,
+      "step": 953500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5038878224940627e-06,
+      "loss": 0.8062,
+      "step": 954000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4784705729251392e-06,
+      "loss": 0.8002,
+      "step": 954500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.4530533233562156e-06,
+      "loss": 0.7978,
+      "step": 955000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4276360737872923e-06,
+      "loss": 0.8154,
+      "step": 955500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4022188242183689e-06,
+      "loss": 0.827,
+      "step": 956000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3768015746494455e-06,
+      "loss": 0.8179,
+      "step": 956500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3513843250805218e-06,
+      "loss": 0.8273,
+      "step": 957000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3259670755115986e-06,
+      "loss": 0.8126,
+      "step": 957500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3005498259426751e-06,
+      "loss": 0.7991,
+      "step": 958000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2751325763737515e-06,
+      "loss": 0.8078,
+      "step": 958500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.249715326804828e-06,
+      "loss": 0.7757,
+      "step": 959000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2242980772359046e-06,
+      "loss": 0.8349,
+      "step": 959500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1988808276669814e-06,
+      "loss": 0.8088,
+      "step": 960000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1734635780980577e-06,
+      "loss": 0.8077,
+      "step": 960500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1480463285291343e-06,
+      "loss": 0.8285,
+      "step": 961000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1226290789602108e-06,
+      "loss": 0.8014,
+      "step": 961500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0972118293912876e-06,
+      "loss": 0.8111,
+      "step": 962000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.071794579822364e-06,
+      "loss": 0.799,
+      "step": 962500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0463773302534405e-06,
+      "loss": 0.7945,
+      "step": 963000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.020960080684517e-06,
+      "loss": 0.808,
+      "step": 963500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.955428311155936e-07,
+      "loss": 0.8083,
+      "step": 964000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.701255815466702e-07,
+      "loss": 0.8094,
+      "step": 964500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.447083319777467e-07,
+      "loss": 0.8373,
+      "step": 965000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.192910824088233e-07,
+      "loss": 0.8204,
+      "step": 965500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.938738328398997e-07,
+      "loss": 0.7997,
+      "step": 966000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.684565832709764e-07,
+      "loss": 0.8268,
+      "step": 966500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.43039333702053e-07,
+      "loss": 0.8135,
+      "step": 967000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.176220841331294e-07,
+      "loss": 0.8216,
+      "step": 967500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.92204834564206e-07,
+      "loss": 0.8218,
+      "step": 968000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.667875849952825e-07,
+      "loss": 0.7883,
+      "step": 968500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.413703354263592e-07,
+      "loss": 0.8333,
+      "step": 969000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.159530858574356e-07,
+      "loss": 0.8003,
+      "step": 969500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.905358362885123e-07,
+      "loss": 0.8152,
+      "step": 970000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.651185867195888e-07,
+      "loss": 0.822,
+      "step": 970500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.397013371506654e-07,
+      "loss": 0.8225,
+      "step": 971000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.142840875817419e-07,
+      "loss": 0.8322,
+      "step": 971500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.888668380128185e-07,
+      "loss": 0.8178,
+      "step": 972000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.63449588443895e-07,
+      "loss": 0.8364,
+      "step": 972500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.380323388749716e-07,
+      "loss": 0.8232,
+      "step": 973000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.126150893060481e-07,
+      "loss": 0.8085,
+      "step": 973500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.871978397371247e-07,
+      "loss": 0.8405,
+      "step": 974000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.6178059016820123e-07,
+      "loss": 0.8252,
+      "step": 974500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3636334059927773e-07,
+      "loss": 0.8038,
+      "step": 975000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1094609103035434e-07,
+      "loss": 0.8295,
+      "step": 975500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.8552884146143085e-07,
+      "loss": 0.8056,
+      "step": 976000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.601115918925074e-07,
+      "loss": 0.8043,
+      "step": 976500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.3469434232358397e-07,
+      "loss": 0.7801,
+      "step": 977000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.092770927546605e-07,
+      "loss": 0.8496,
+      "step": 977500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.838598431857371e-07,
+      "loss": 0.7923,
+      "step": 978000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.5844259361681364e-07,
+      "loss": 0.8031,
+      "step": 978500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.3302534404789017e-07,
+      "loss": 0.801,
+      "step": 979000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0760809447896673e-07,
+      "loss": 0.8196,
+      "step": 979500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.821908449100433e-07,
+      "loss": 0.8259,
+      "step": 980000
+    }
+  ],
+  "max_steps": 983584,
+  "num_train_epochs": 4,
+  "total_flos": 4.118080930442403e+18,
+  "trial_name": null,
+  "trial_params": null
+}