diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6574 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.939373084926839,
+  "eval_steps": 1000000,
+  "global_step": 545561,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 10.3042,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 10.1332,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 10.0253,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 9.9113,
+      "step": 2000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5e-06,
+      "loss": 9.7712,
+      "step": 2500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3e-06,
+      "loss": 9.5994,
+      "step": 3000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.5e-06,
+      "loss": 9.3977,
+      "step": 3500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 9.1727,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.5e-06,
+      "loss": 8.9273,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-06,
+      "loss": 8.676,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 8.4202,
+      "step": 5500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6e-06,
+      "loss": 8.1709,
+      "step": 6000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 7.9386,
+      "step": 6500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7e-06,
+      "loss": 7.7288,
+      "step": 7000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 7.5463,
+      "step": 7500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 7.402,
+      "step": 8000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.5e-06,
+      "loss": 7.2978,
+      "step": 8500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9e-06,
+      "loss": 7.2228,
+      "step": 9000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5e-06,
+      "loss": 7.1618,
+      "step": 9500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1e-05,
+      "loss": 7.1108,
+      "step": 10000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.999179372712501e-06,
+      "loss": 7.0682,
+      "step": 10500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.998358745425003e-06,
+      "loss": 7.0207,
+      "step": 11000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.997538118137506e-06,
+      "loss": 6.9944,
+      "step": 11500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.996717490850006e-06,
+      "loss": 6.9651,
+      "step": 12000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.995896863562508e-06,
+      "loss": 6.9362,
+      "step": 12500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.99507623627501e-06,
+      "loss": 6.9162,
+      "step": 13000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.994255608987511e-06,
+      "loss": 6.8912,
+      "step": 13500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.993434981700011e-06,
+      "loss": 6.8696,
+      "step": 14000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.992614354412514e-06,
+      "loss": 6.8525,
+      "step": 14500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.991793727125016e-06,
+      "loss": 6.8344,
+      "step": 15000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.990973099837516e-06,
+      "loss": 6.8146,
+      "step": 15500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.990152472550018e-06,
+      "loss": 6.8026,
+      "step": 16000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.98933184526252e-06,
+      "loss": 6.7849,
+      "step": 16500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.988511217975021e-06,
+      "loss": 6.7637,
+      "step": 17000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.987690590687521e-06,
+      "loss": 6.7575,
+      "step": 17500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.986869963400024e-06,
+      "loss": 6.7497,
+      "step": 18000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.986049336112526e-06,
+      "loss": 6.7278,
+      "step": 18500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.985228708825026e-06,
+      "loss": 6.7218,
+      "step": 19000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.984408081537528e-06,
+      "loss": 6.7127,
+      "step": 19500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.98358745425003e-06,
+      "loss": 6.697,
+      "step": 20000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.982766826962531e-06,
+      "loss": 6.6915,
+      "step": 20500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.981946199675032e-06,
+      "loss": 6.6818,
+      "step": 21000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.981125572387534e-06,
+      "loss": 6.6712,
+      "step": 21500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.980304945100036e-06,
+      "loss": 6.6654,
+      "step": 22000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.979484317812536e-06,
+      "loss": 6.6577,
+      "step": 22500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.978663690525039e-06,
+      "loss": 6.6435,
+      "step": 23000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.97784306323754e-06,
+      "loss": 6.6364,
+      "step": 23500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.977022435950041e-06,
+      "loss": 6.6359,
+      "step": 24000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.976201808662542e-06,
+      "loss": 6.6209,
+      "step": 24500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.975381181375044e-06,
+      "loss": 6.6216,
+      "step": 25000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.974560554087546e-06,
+      "loss": 6.6126,
+      "step": 25500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.973739926800046e-06,
+      "loss": 6.6012,
+      "step": 26000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.972919299512549e-06,
+      "loss": 6.5936,
+      "step": 26500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.97209867222505e-06,
+      "loss": 6.59,
+      "step": 27000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.971278044937551e-06,
+      "loss": 6.5814,
+      "step": 27500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.970457417650052e-06,
+      "loss": 6.577,
+      "step": 28000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.969636790362554e-06,
+      "loss": 6.5701,
+      "step": 28500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.968816163075056e-06,
+      "loss": 6.5651,
+      "step": 29000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.967995535787557e-06,
+      "loss": 6.5587,
+      "step": 29500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.967174908500059e-06,
+      "loss": 6.5533,
+      "step": 30000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.966354281212561e-06,
+      "loss": 6.5508,
+      "step": 30500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.965533653925061e-06,
+      "loss": 6.5454,
+      "step": 31000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964713026637562e-06,
+      "loss": 6.5347,
+      "step": 31500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963892399350064e-06,
+      "loss": 6.5301,
+      "step": 32000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.963071772062566e-06,
+      "loss": 6.5268,
+      "step": 32500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.962251144775067e-06,
+      "loss": 6.5315,
+      "step": 33000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.961430517487569e-06,
+      "loss": 6.5217,
+      "step": 33500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.960609890200071e-06,
+      "loss": 6.5108,
+      "step": 34000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.959789262912571e-06,
+      "loss": 6.5101,
+      "step": 34500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.958968635625072e-06,
+      "loss": 6.5011,
+      "step": 35000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.958148008337574e-06,
+      "loss": 6.5033,
+      "step": 35500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.957327381050076e-06,
+      "loss": 6.4988,
+      "step": 36000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.956506753762577e-06,
+      "loss": 6.4913,
+      "step": 36500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.955686126475079e-06,
+      "loss": 6.486,
+      "step": 37000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.95486549918758e-06,
+      "loss": 6.4853,
+      "step": 37500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.954044871900081e-06,
+      "loss": 6.4824,
+      "step": 38000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.953224244612582e-06,
+      "loss": 6.477,
+      "step": 38500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.952403617325084e-06,
+      "loss": 6.4701,
+      "step": 39000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.951582990037586e-06,
+      "loss": 6.4647,
+      "step": 39500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.950762362750087e-06,
+      "loss": 6.4659,
+      "step": 40000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.949941735462589e-06,
+      "loss": 6.459,
+      "step": 40500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.94912110817509e-06,
+      "loss": 6.4549,
+      "step": 41000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.948300480887592e-06,
+      "loss": 6.4515,
+      "step": 41500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.947479853600092e-06,
+      "loss": 6.447,
+      "step": 42000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.946659226312594e-06,
+      "loss": 6.4387,
+      "step": 42500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.945838599025096e-06,
+      "loss": 6.4397,
+      "step": 43000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.945017971737597e-06,
+      "loss": 6.4298,
+      "step": 43500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.944197344450099e-06,
+      "loss": 6.4316,
+      "step": 44000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.9433767171626e-06,
+      "loss": 6.4274,
+      "step": 44500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.942556089875102e-06,
+      "loss": 6.4252,
+      "step": 45000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.941735462587602e-06,
+      "loss": 6.4199,
+      "step": 45500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.940914835300104e-06,
+      "loss": 6.4161,
+      "step": 46000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.940094208012606e-06,
+      "loss": 6.4129,
+      "step": 46500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.939273580725107e-06,
+      "loss": 6.4131,
+      "step": 47000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.938452953437609e-06,
+      "loss": 6.4028,
+      "step": 47500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.93763232615011e-06,
+      "loss": 6.408,
+      "step": 48000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.936811698862612e-06,
+      "loss": 6.401,
+      "step": 48500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.935991071575112e-06,
+      "loss": 6.3963,
+      "step": 49000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.935170444287614e-06,
+      "loss": 6.3993,
+      "step": 49500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.934349817000117e-06,
+      "loss": 6.3973,
+      "step": 50000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.933529189712617e-06,
+      "loss": 6.3868,
+      "step": 50500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.93270856242512e-06,
+      "loss": 6.3873,
+      "step": 51000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.93188793513762e-06,
+      "loss": 6.3871,
+      "step": 51500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.931067307850122e-06,
+      "loss": 6.3834,
+      "step": 52000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.930246680562622e-06,
+      "loss": 6.3719,
+      "step": 52500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.929426053275124e-06,
+      "loss": 6.375,
+      "step": 53000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.928605425987625e-06,
+      "loss": 6.3809,
+      "step": 53500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.927784798700127e-06,
+      "loss": 6.371,
+      "step": 54000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.92696417141263e-06,
+      "loss": 6.3689,
+      "step": 54500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.92614354412513e-06,
+      "loss": 6.3661,
+      "step": 55000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.925322916837632e-06,
+      "loss": 6.3637,
+      "step": 55500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.924502289550132e-06,
+      "loss": 6.3668,
+      "step": 56000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.923681662262635e-06,
+      "loss": 6.3583,
+      "step": 56500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.922861034975135e-06,
+      "loss": 6.3629,
+      "step": 57000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.922040407687637e-06,
+      "loss": 6.3546,
+      "step": 57500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.92121978040014e-06,
+      "loss": 6.3567,
+      "step": 58000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.92039915311264e-06,
+      "loss": 6.3511,
+      "step": 58500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.919578525825142e-06,
+      "loss": 6.3437,
+      "step": 59000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.918757898537642e-06,
+      "loss": 6.3478,
+      "step": 59500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.917937271250145e-06,
+      "loss": 6.3438,
+      "step": 60000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.917116643962645e-06,
+      "loss": 6.3399,
+      "step": 60500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.916296016675147e-06,
+      "loss": 6.3414,
+      "step": 61000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.91547538938765e-06,
+      "loss": 6.3309,
+      "step": 61500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.91465476210015e-06,
+      "loss": 6.3386,
+      "step": 62000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.913834134812652e-06,
+      "loss": 6.3336,
+      "step": 62500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.913013507525153e-06,
+      "loss": 6.3314,
+      "step": 63000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.912192880237655e-06,
+      "loss": 6.3303,
+      "step": 63500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.911372252950155e-06,
+      "loss": 6.3238,
+      "step": 64000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.910551625662657e-06,
+      "loss": 6.32,
+      "step": 64500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.90973099837516e-06,
+      "loss": 6.3247,
+      "step": 65000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.90891037108766e-06,
+      "loss": 6.3222,
+      "step": 65500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.908089743800162e-06,
+      "loss": 6.3203,
+      "step": 66000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.907269116512663e-06,
+      "loss": 6.3157,
+      "step": 66500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.906448489225165e-06,
+      "loss": 6.3152,
+      "step": 67000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.905627861937665e-06,
+      "loss": 6.3119,
+      "step": 67500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.904807234650167e-06,
+      "loss": 6.3095,
+      "step": 68000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.90398660736267e-06,
+      "loss": 6.3103,
+      "step": 68500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.90316598007517e-06,
+      "loss": 6.3061,
+      "step": 69000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.90234535278767e-06,
+      "loss": 6.3076,
+      "step": 69500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.901524725500173e-06,
+      "loss": 6.306,
+      "step": 70000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.900704098212675e-06,
+      "loss": 6.3018,
+      "step": 70500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.899883470925175e-06,
+      "loss": 6.296,
+      "step": 71000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.899062843637678e-06,
+      "loss": 6.2994,
+      "step": 71500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.89824221635018e-06,
+      "loss": 6.297,
+      "step": 72000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.89742158906268e-06,
+      "loss": 6.2908,
+      "step": 72500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.89660096177518e-06,
+      "loss": 6.2935,
+      "step": 73000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.895780334487683e-06,
+      "loss": 6.2881,
+      "step": 73500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.894959707200185e-06,
+      "loss": 6.2873,
+      "step": 74000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.894139079912685e-06,
+      "loss": 6.2872,
+      "step": 74500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.893318452625188e-06,
+      "loss": 6.2814,
+      "step": 75000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.89249782533769e-06,
+      "loss": 6.2823,
+      "step": 75500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.89167719805019e-06,
+      "loss": 6.2814,
+      "step": 76000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.89085657076269e-06,
+      "loss": 6.283,
+      "step": 76500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.890035943475193e-06,
+      "loss": 6.2811,
+      "step": 77000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.889215316187695e-06,
+      "loss": 6.278,
+      "step": 77500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.888394688900196e-06,
+      "loss": 6.2794,
+      "step": 78000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.887574061612698e-06,
+      "loss": 6.2761,
+      "step": 78500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.8867534343252e-06,
+      "loss": 6.2801,
+      "step": 79000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.8859328070377e-06,
+      "loss": 6.2675,
+      "step": 79500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.8851121797502e-06,
+      "loss": 6.2741,
+      "step": 80000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.884291552462703e-06,
+      "loss": 6.2732,
+      "step": 80500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.883470925175205e-06,
+      "loss": 6.268,
+      "step": 81000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.882650297887706e-06,
+      "loss": 6.2676,
+      "step": 81500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.881829670600208e-06,
+      "loss": 6.2609,
+      "step": 82000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.88100904331271e-06,
+      "loss": 6.2648,
+      "step": 82500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.88018841602521e-06,
+      "loss": 6.2615,
+      "step": 83000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.879367788737711e-06,
+      "loss": 6.2625,
+      "step": 83500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.878547161450213e-06,
+      "loss": 6.2595,
+      "step": 84000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.877726534162715e-06,
+      "loss": 6.2516,
+      "step": 84500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.876905906875216e-06,
+      "loss": 6.2633,
+      "step": 85000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.876085279587718e-06,
+      "loss": 6.2589,
+      "step": 85500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.87526465230022e-06,
+      "loss": 6.2546,
+      "step": 86000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.87444402501272e-06,
+      "loss": 6.2542,
+      "step": 86500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.873623397725221e-06,
+      "loss": 6.2507,
+      "step": 87000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.872802770437723e-06,
+      "loss": 6.2537,
+      "step": 87500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.871982143150225e-06,
+      "loss": 6.2563,
+      "step": 88000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.871161515862726e-06,
+      "loss": 6.2486,
+      "step": 88500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.870340888575228e-06,
+      "loss": 6.2487,
+      "step": 89000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.86952026128773e-06,
+      "loss": 6.2452,
+      "step": 89500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.86869963400023e-06,
+      "loss": 6.2454,
+      "step": 90000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.867879006712731e-06,
+      "loss": 6.2442,
+      "step": 90500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.867058379425233e-06,
+      "loss": 6.2388,
+      "step": 91000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.866237752137735e-06,
+      "loss": 6.2386,
+      "step": 91500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.865417124850236e-06,
+      "loss": 6.2399,
+      "step": 92000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.864596497562738e-06,
+      "loss": 6.2369,
+      "step": 92500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.86377587027524e-06,
+      "loss": 6.2317,
+      "step": 93000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.86295524298774e-06,
+      "loss": 6.2344,
+      "step": 93500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.862134615700241e-06,
+      "loss": 6.2373,
+      "step": 94000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.861313988412743e-06,
+      "loss": 6.2386,
+      "step": 94500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.860493361125245e-06,
+      "loss": 6.2394,
+      "step": 95000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.859672733837746e-06,
+      "loss": 6.2314,
+      "step": 95500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.858852106550248e-06,
+      "loss": 6.23,
+      "step": 96000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.85803147926275e-06,
+      "loss": 6.2324,
+      "step": 96500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.85721085197525e-06,
+      "loss": 6.2309,
+      "step": 97000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.856390224687751e-06,
+      "loss": 6.2301,
+      "step": 97500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.855569597400253e-06,
+      "loss": 6.2292,
+      "step": 98000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.854748970112756e-06,
+      "loss": 6.2338,
+      "step": 98500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.853928342825256e-06,
+      "loss": 6.2243,
+      "step": 99000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.853107715537758e-06,
+      "loss": 6.2216,
+      "step": 99500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.85228708825026e-06,
+      "loss": 6.2295,
+      "step": 100000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.851466460962761e-06,
+      "loss": 6.2231,
+      "step": 100500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.850645833675261e-06,
+      "loss": 6.224,
+      "step": 101000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.849825206387764e-06,
+      "loss": 6.2211,
+      "step": 101500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.849004579100266e-06,
+      "loss": 6.2171,
+      "step": 102000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.848183951812766e-06,
+      "loss": 6.2208,
+      "step": 102500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.847363324525268e-06,
+      "loss": 6.2197,
+      "step": 103000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.84654269723777e-06,
+      "loss": 6.2113,
+      "step": 103500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.845722069950271e-06,
+      "loss": 6.2144,
+      "step": 104000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.844901442662771e-06,
+      "loss": 6.2182,
+      "step": 104500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.844080815375274e-06,
+      "loss": 6.2148,
+      "step": 105000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.843260188087776e-06,
+      "loss": 6.2136,
+      "step": 105500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.842439560800276e-06,
+      "loss": 6.2133,
+      "step": 106000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.841618933512778e-06,
+      "loss": 6.2133,
+      "step": 106500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.84079830622528e-06,
+      "loss": 6.2141,
+      "step": 107000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.839977678937781e-06,
+      "loss": 6.2118,
+      "step": 107500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.839157051650282e-06,
+      "loss": 6.2084,
+      "step": 108000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.838336424362784e-06,
+      "loss": 6.2077,
+      "step": 108500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.837515797075286e-06,
+      "loss": 6.2059,
+      "step": 109000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.836695169787786e-06,
+      "loss": 6.2038,
+      "step": 109500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.835874542500288e-06,
+      "loss": 6.2015,
+      "step": 110000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.83505391521279e-06,
+      "loss": 6.2063,
+      "step": 110500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.834233287925291e-06,
+      "loss": 6.203,
+      "step": 111000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.833412660637792e-06,
+      "loss": 6.2064,
+      "step": 111500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.832592033350294e-06,
+      "loss": 6.2053,
+      "step": 112000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.831771406062796e-06,
+      "loss": 6.205,
+      "step": 112500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.830950778775296e-06,
+      "loss": 6.203,
+      "step": 113000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.830130151487799e-06,
+      "loss": 6.2047,
+      "step": 113500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.8293095242003e-06,
+      "loss": 6.2026,
+      "step": 114000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.828488896912801e-06,
+      "loss": 6.1965,
+      "step": 114500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.827668269625302e-06,
+      "loss": 6.1989,
+      "step": 115000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.826847642337804e-06,
+      "loss": 6.1943,
+      "step": 115500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.826027015050306e-06,
+      "loss": 6.198,
+      "step": 116000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.825206387762806e-06,
+      "loss": 6.1928,
+      "step": 116500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.824385760475309e-06,
+      "loss": 6.1972,
+      "step": 117000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.823565133187809e-06,
+      "loss": 6.1888,
+      "step": 117500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.822744505900311e-06,
+      "loss": 6.1961,
+      "step": 118000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.821923878612812e-06,
+      "loss": 6.1909,
+      "step": 118500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.821103251325314e-06,
+      "loss": 6.1956,
+      "step": 119000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.820282624037816e-06,
+      "loss": 6.1922,
+      "step": 119500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.819461996750317e-06,
+      "loss": 6.1916,
+      "step": 120000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.818641369462819e-06,
+      "loss": 6.1864,
+      "step": 120500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.81782074217532e-06,
+      "loss": 6.189,
+      "step": 121000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.817000114887821e-06,
+      "loss": 6.1866,
+      "step": 121500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.816179487600322e-06,
+      "loss": 6.188,
+      "step": 122000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.815358860312824e-06,
+      "loss": 6.1868,
+      "step": 122500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.814538233025326e-06,
+      "loss": 6.1843,
+      "step": 123000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.813717605737827e-06,
+      "loss": 6.1842,
+      "step": 123500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.812896978450329e-06,
+      "loss": 6.1844,
+      "step": 124000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.81207635116283e-06,
+      "loss": 6.1856,
+      "step": 124500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.811255723875331e-06,
+      "loss": 6.1796,
+      "step": 125000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.810435096587832e-06,
+      "loss": 6.1803,
+      "step": 125500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.809614469300334e-06,
+      "loss": 6.1792,
+      "step": 126000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.808793842012836e-06,
+      "loss": 6.1835,
+      "step": 126500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.807973214725337e-06,
+      "loss": 6.175,
+      "step": 127000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.807152587437839e-06,
+      "loss": 6.183,
+      "step": 127500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.80633196015034e-06,
+      "loss": 6.1779,
+      "step": 128000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.805511332862842e-06,
+      "loss": 6.1803,
+      "step": 128500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.804690705575342e-06,
+      "loss": 6.1787,
+      "step": 129000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.803870078287844e-06,
+      "loss": 6.1772,
+      "step": 129500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.803049451000346e-06,
+      "loss": 6.1762,
+      "step": 130000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.802228823712847e-06,
+      "loss": 6.1758,
+      "step": 130500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.801408196425349e-06,
+      "loss": 6.1752,
+      "step": 131000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.80058756913785e-06,
+      "loss": 6.1765,
+      "step": 131500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.799766941850352e-06,
+      "loss": 6.1774,
+      "step": 132000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.798946314562852e-06,
+      "loss": 6.1771,
+      "step": 132500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.798125687275354e-06,
+      "loss": 6.168,
+      "step": 133000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.797305059987855e-06,
+      "loss": 6.1753,
+      "step": 133500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.796484432700357e-06,
+      "loss": 6.1712,
+      "step": 134000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.795663805412859e-06,
+      "loss": 6.1702,
+      "step": 134500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.79484317812536e-06,
+      "loss": 6.1725,
+      "step": 135000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.794022550837862e-06,
+      "loss": 6.1724,
+      "step": 135500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.793201923550362e-06,
+      "loss": 6.1697,
+      "step": 136000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.792381296262864e-06,
+      "loss": 6.1658,
+      "step": 136500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.791560668975365e-06,
+      "loss": 6.1638,
+      "step": 137000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.790740041687867e-06,
+      "loss": 6.1648,
+      "step": 137500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.78991941440037e-06,
+      "loss": 6.1654,
+      "step": 138000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.78909878711287e-06,
+      "loss": 6.1614,
+      "step": 138500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.788278159825372e-06,
+      "loss": 6.1715,
+      "step": 139000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.787457532537872e-06,
+      "loss": 6.1657,
+      "step": 139500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.786636905250374e-06,
+      "loss": 6.1651,
+      "step": 140000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.785816277962875e-06,
+      "loss": 6.1645,
+      "step": 140500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.784995650675377e-06,
+      "loss": 6.1671,
+      "step": 141000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.78417502338788e-06,
+      "loss": 6.1601,
+      "step": 141500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.78335439610038e-06,
+      "loss": 6.1605,
+      "step": 142000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.782533768812882e-06,
+      "loss": 6.1619,
+      "step": 142500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.781713141525382e-06,
+      "loss": 6.1659,
+      "step": 143000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.780892514237885e-06,
+      "loss": 6.1572,
+      "step": 143500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.780071886950385e-06,
+      "loss": 6.1578,
+      "step": 144000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.779251259662887e-06,
+      "loss": 6.1552,
+      "step": 144500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.77843063237539e-06,
+      "loss": 6.1603,
+      "step": 145000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.77761000508789e-06,
+      "loss": 6.1594,
+      "step": 145500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.776789377800392e-06,
+      "loss": 6.1564,
+      "step": 146000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.775968750512892e-06,
+      "loss": 6.1606,
+      "step": 146500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.775148123225395e-06,
+      "loss": 6.1563,
+      "step": 147000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.774327495937895e-06,
+      "loss": 6.1568,
+      "step": 147500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.773506868650397e-06,
+      "loss": 6.1504,
+      "step": 148000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.7726862413629e-06,
+      "loss": 6.1572,
+      "step": 148500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.7718656140754e-06,
+      "loss": 6.1515,
+      "step": 149000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.7710449867879e-06,
+      "loss": 6.1491,
+      "step": 149500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.770224359500403e-06,
+      "loss": 6.1542,
+      "step": 150000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.769403732212905e-06,
+      "loss": 6.15,
+      "step": 150500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.768583104925405e-06,
+      "loss": 6.1529,
+      "step": 151000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.767762477637907e-06,
+      "loss": 6.1561,
+      "step": 151500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.76694185035041e-06,
+      "loss": 6.1457,
+      "step": 152000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.76612122306291e-06,
+      "loss": 6.1502,
+      "step": 152500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.76530059577541e-06,
+      "loss": 6.1462,
+      "step": 153000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.764479968487913e-06,
+      "loss": 6.1474,
+      "step": 153500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.763659341200415e-06,
+      "loss": 6.1421,
+      "step": 154000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.762838713912915e-06,
+      "loss": 6.1476,
+      "step": 154500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.762018086625417e-06,
+      "loss": 6.1477,
+      "step": 155000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.76119745933792e-06,
+      "loss": 6.1442,
+      "step": 155500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.76037683205042e-06,
+      "loss": 6.1429,
+      "step": 156000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.75955620476292e-06,
+      "loss": 6.1436,
+      "step": 156500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.758735577475423e-06,
+      "loss": 6.1503,
+      "step": 157000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.757914950187925e-06,
+      "loss": 6.1471,
+      "step": 157500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.757094322900425e-06,
+      "loss": 6.147,
+      "step": 158000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.756273695612928e-06,
+      "loss": 6.1433,
+      "step": 158500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75545306832543e-06,
+      "loss": 6.1399,
+      "step": 159000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75463244103793e-06,
+      "loss": 6.1444,
+      "step": 159500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.75381181375043e-06,
+      "loss": 6.1422,
+      "step": 160000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.752991186462933e-06,
+      "loss": 6.1441,
+      "step": 160500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.752170559175435e-06,
+      "loss": 6.1367,
+      "step": 161000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.751349931887935e-06,
+      "loss": 6.1403,
+      "step": 161500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.750529304600438e-06,
+      "loss": 6.1392,
+      "step": 162000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.74970867731294e-06,
+      "loss": 6.1353,
+      "step": 162500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.74888805002544e-06,
+      "loss": 6.1393,
+      "step": 163000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.74806742273794e-06,
+      "loss": 6.1415,
+      "step": 163500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.747246795450443e-06,
+      "loss": 6.1392,
+      "step": 164000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.746426168162945e-06,
+      "loss": 6.1379,
+      "step": 164500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.745605540875446e-06,
+      "loss": 6.1393,
+      "step": 165000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.744784913587948e-06,
+      "loss": 6.1379,
+      "step": 165500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.74396428630045e-06,
+      "loss": 6.1345,
+      "step": 166000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.74314365901295e-06,
+      "loss": 6.1405,
+      "step": 166500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.74232303172545e-06,
+      "loss": 6.1339,
+      "step": 167000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.741502404437953e-06,
+      "loss": 6.1347,
+      "step": 167500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.740681777150455e-06,
+      "loss": 6.1351,
+      "step": 168000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.739861149862956e-06,
+      "loss": 6.1284,
+      "step": 168500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.739040522575458e-06,
+      "loss": 6.1294,
+      "step": 169000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.73821989528796e-06,
+      "loss": 6.1372,
+      "step": 169500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.73739926800046e-06,
+      "loss": 6.1343,
+      "step": 170000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.736578640712961e-06,
+      "loss": 6.1312,
+      "step": 170500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.735758013425463e-06,
+      "loss": 6.132,
+      "step": 171000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.734937386137965e-06,
+      "loss": 6.1358,
+      "step": 171500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.734116758850466e-06,
+      "loss": 6.1296,
+      "step": 172000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.733296131562968e-06,
+      "loss": 6.132,
+      "step": 172500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.73247550427547e-06,
+      "loss": 6.1295,
+      "step": 173000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.73165487698797e-06,
+      "loss": 6.129,
+      "step": 173500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.730834249700471e-06,
+      "loss": 6.1301,
+      "step": 174000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.730013622412973e-06,
+      "loss": 6.1264,
+      "step": 174500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.729192995125475e-06,
+      "loss": 6.1298,
+      "step": 175000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.728372367837976e-06,
+      "loss": 6.1248,
+      "step": 175500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.727551740550478e-06,
+      "loss": 6.1232,
+      "step": 176000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.72673111326298e-06,
+      "loss": 6.1249,
+      "step": 176500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.72591048597548e-06,
+      "loss": 6.1288,
+      "step": 177000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.725089858687981e-06,
+      "loss": 6.1234,
+      "step": 177500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.724269231400483e-06,
+      "loss": 6.1232,
+      "step": 178000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.723448604112985e-06,
+      "loss": 6.1328,
+      "step": 178500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.722627976825486e-06,
+      "loss": 6.1245,
+      "step": 179000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.721807349537988e-06,
+      "loss": 6.1203,
+      "step": 179500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.72098672225049e-06,
+      "loss": 6.1264,
+      "step": 180000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.72016609496299e-06,
+      "loss": 6.123,
+      "step": 180500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.719345467675491e-06,
+      "loss": 6.1273,
+      "step": 181000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.718524840387993e-06,
+      "loss": 6.1209,
+      "step": 181500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.717704213100495e-06,
+      "loss": 6.123,
+      "step": 182000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.716883585812996e-06,
+      "loss": 6.123,
+      "step": 182500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.716062958525498e-06,
+      "loss": 6.1178,
+      "step": 183000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.715242331238e-06,
+      "loss": 6.1241,
+      "step": 183500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.7144217039505e-06,
+      "loss": 6.1201,
+      "step": 184000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.713601076663001e-06,
+      "loss": 6.1177,
+      "step": 184500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.712780449375503e-06,
+      "loss": 6.1227,
+      "step": 185000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.711959822088006e-06,
+      "loss": 6.1188,
+      "step": 185500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.711139194800506e-06,
+      "loss": 6.1176,
+      "step": 186000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.710318567513008e-06,
+      "loss": 6.1189,
+      "step": 186500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.70949794022551e-06,
+      "loss": 6.1173,
+      "step": 187000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.70867731293801e-06,
+      "loss": 6.1212,
+      "step": 187500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.707856685650511e-06,
+      "loss": 6.1182,
+      "step": 188000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.707036058363013e-06,
+      "loss": 6.1156,
+      "step": 188500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.706215431075516e-06,
+      "loss": 6.1206,
+      "step": 189000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.705394803788016e-06,
+      "loss": 6.1133,
+      "step": 189500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.704574176500518e-06,
+      "loss": 6.1175,
+      "step": 190000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.70375354921302e-06,
+      "loss": 6.1203,
+      "step": 190500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.702932921925521e-06,
+      "loss": 6.1109,
+      "step": 191000
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.702112294638021e-06,
+      "loss": 6.1088,
+      "step": 191500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.701291667350524e-06,
+      "loss": 6.1127,
+      "step": 192000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.700471040063026e-06,
+      "loss": 6.1116,
+      "step": 192500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.699650412775526e-06,
+      "loss": 6.1163,
+      "step": 193000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.698829785488028e-06,
+      "loss": 6.1151,
+      "step": 193500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.69800915820053e-06,
+      "loss": 6.1138,
+      "step": 194000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.697188530913031e-06,
+      "loss": 6.1159,
+      "step": 194500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.696367903625531e-06,
+      "loss": 6.1104,
+      "step": 195000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.695547276338034e-06,
+      "loss": 6.1148,
+      "step": 195500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.694726649050536e-06,
+      "loss": 6.1123,
+      "step": 196000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.693906021763036e-06,
+      "loss": 6.1129,
+      "step": 196500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.693085394475538e-06,
+      "loss": 6.1064,
+      "step": 197000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.69226476718804e-06,
+      "loss": 6.1094,
+      "step": 197500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.691444139900541e-06,
+      "loss": 6.1114,
+      "step": 198000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.690623512613042e-06,
+      "loss": 6.1122,
+      "step": 198500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.689802885325544e-06,
+      "loss": 6.1139,
+      "step": 199000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.688982258038046e-06,
+      "loss": 6.1083,
+      "step": 199500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.688161630750546e-06,
+      "loss": 6.1048,
+      "step": 200000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.687341003463049e-06,
+      "loss": 6.1086,
+      "step": 200500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.686520376175549e-06,
+      "loss": 6.1088,
+      "step": 201000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.685699748888051e-06,
+      "loss": 6.1056,
+      "step": 201500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.684879121600552e-06,
+      "loss": 6.1083,
+      "step": 202000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.684058494313054e-06,
+      "loss": 6.1095,
+      "step": 202500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.683237867025556e-06,
+      "loss": 6.1076,
+      "step": 203000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.682417239738056e-06,
+      "loss": 6.1086,
+      "step": 203500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.681596612450559e-06,
+      "loss": 6.1015,
+      "step": 204000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.680775985163059e-06,
+      "loss": 6.1035,
+      "step": 204500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.679955357875561e-06,
+      "loss": 6.1082,
+      "step": 205000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.679134730588062e-06,
+      "loss": 6.1025,
+      "step": 205500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.678314103300564e-06,
+      "loss": 6.1041,
+      "step": 206000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.677493476013066e-06,
+      "loss": 6.0991,
+      "step": 206500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.676672848725567e-06,
+      "loss": 6.1055,
+      "step": 207000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.675852221438069e-06,
+      "loss": 6.1053,
+      "step": 207500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.67503159415057e-06,
+      "loss": 6.1059,
+      "step": 208000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.674210966863071e-06,
+      "loss": 6.0999,
+      "step": 208500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.673390339575572e-06,
+      "loss": 6.1035,
+      "step": 209000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.672569712288074e-06,
+      "loss": 6.0976,
+      "step": 209500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.671749085000576e-06,
+      "loss": 6.1043,
+      "step": 210000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.670928457713077e-06,
+      "loss": 6.1046,
+      "step": 210500
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.670107830425579e-06,
+      "loss": 6.0997,
+      "step": 211000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.66928720313808e-06,
+      "loss": 6.0964,
+      "step": 211500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.668466575850581e-06,
+      "loss": 6.1018,
+      "step": 212000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.667645948563082e-06,
+      "loss": 6.1007,
+      "step": 212500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.666825321275584e-06,
+      "loss": 6.1017,
+      "step": 213000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.666004693988086e-06,
+      "loss": 6.0981,
+      "step": 213500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.665184066700587e-06,
+      "loss": 6.0978,
+      "step": 214000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.664363439413089e-06,
+      "loss": 6.0975,
+      "step": 214500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.66354281212559e-06,
+      "loss": 6.1039,
+      "step": 215000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 9.662722184838092e-06,
+      "loss": 6.0982,
+      "step": 215500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.661901557550592e-06,
+      "loss": 6.0972,
+      "step": 216000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 9.661080930263094e-06,
+      "loss": 6.0984,
+      "step": 216500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.660260302975595e-06,
+      "loss": 6.1005,
+      "step": 217000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.659439675688097e-06,
+      "loss": 6.102,
+      "step": 217500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.658619048400599e-06,
+      "loss": 6.0925,
+      "step": 218000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.6577984211131e-06,
+      "loss": 6.0964,
+      "step": 218500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.656977793825602e-06,
+      "loss": 6.0967,
+      "step": 219000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.656157166538102e-06,
+      "loss": 6.0957,
+      "step": 219500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.655336539250604e-06,
+      "loss": 6.0949,
+      "step": 220000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.654515911963105e-06,
+      "loss": 6.0962,
+      "step": 220500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.653695284675607e-06,
+      "loss": 6.0881,
+      "step": 221000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.652874657388109e-06,
+      "loss": 6.099,
+      "step": 221500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.65205403010061e-06,
+      "loss": 6.0959,
+      "step": 222000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.651233402813112e-06,
+      "loss": 6.0972,
+      "step": 222500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.650412775525612e-06,
+      "loss": 6.0957,
+      "step": 223000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.649592148238114e-06,
+      "loss": 6.094,
+      "step": 223500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.648771520950615e-06,
+      "loss": 6.0919,
+      "step": 224000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.647950893663117e-06,
+      "loss": 6.0905,
+      "step": 224500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.647130266375619e-06,
+      "loss": 6.0925,
+      "step": 225000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.64630963908812e-06,
+      "loss": 6.0925,
+      "step": 225500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.645489011800622e-06,
+      "loss": 6.0959,
+      "step": 226000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.644668384513122e-06,
+      "loss": 6.091,
+      "step": 226500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.643847757225624e-06,
+      "loss": 6.0928,
+      "step": 227000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.643027129938125e-06,
+      "loss": 6.0979,
+      "step": 227500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.642206502650627e-06,
+      "loss": 6.0936,
+      "step": 228000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.64138587536313e-06,
+      "loss": 6.0927,
+      "step": 228500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.64056524807563e-06,
+      "loss": 6.0919,
+      "step": 229000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.639744620788132e-06,
+      "loss": 6.0932,
+      "step": 229500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.638923993500632e-06,
+      "loss": 6.0895,
+      "step": 230000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.638103366213134e-06,
+      "loss": 6.0891,
+      "step": 230500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.637282738925635e-06,
+      "loss": 6.091,
+      "step": 231000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.636462111638137e-06,
+      "loss": 6.0923,
+      "step": 231500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.63564148435064e-06,
+      "loss": 6.0883,
+      "step": 232000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.63482085706314e-06,
+      "loss": 6.0902,
+      "step": 232500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.63400022977564e-06,
+      "loss": 6.0906,
+      "step": 233000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.633179602488142e-06,
+      "loss": 6.0866,
+      "step": 233500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.632358975200645e-06,
+      "loss": 6.0893,
+      "step": 234000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.631538347913145e-06,
+      "loss": 6.0852,
+      "step": 234500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 9.630717720625647e-06,
+      "loss": 6.0879,
+      "step": 235000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.62989709333815e-06,
+      "loss": 6.0903,
+      "step": 235500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 9.62907646605065e-06,
+      "loss": 6.0886,
+      "step": 236000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.62825583876315e-06,
+      "loss": 6.0847,
+      "step": 236500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.627435211475652e-06,
+      "loss": 6.0872,
+      "step": 237000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.626614584188155e-06,
+      "loss": 6.0894,
+      "step": 237500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 9.625793956900655e-06,
+      "loss": 6.0829,
+      "step": 238000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.624973329613157e-06,
+      "loss": 6.0868,
+      "step": 238500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.62415270232566e-06,
+      "loss": 6.0812,
+      "step": 239000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.62333207503816e-06,
+      "loss": 6.082,
+      "step": 239500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.62251144775066e-06,
+      "loss": 6.0876,
+      "step": 240000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.621690820463163e-06,
+      "loss": 6.0822,
+      "step": 240500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.620870193175665e-06,
+      "loss": 6.0833,
+      "step": 241000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.620049565888165e-06,
+      "loss": 6.0861,
+      "step": 241500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.619228938600667e-06,
+      "loss": 6.085,
+      "step": 242000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.61840831131317e-06,
+      "loss": 6.0851,
+      "step": 242500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.61758768402567e-06,
+      "loss": 6.0839,
+      "step": 243000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.61676705673817e-06,
+      "loss": 6.0864,
+      "step": 243500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.615946429450673e-06,
+      "loss": 6.0804,
+      "step": 244000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.615125802163175e-06,
+      "loss": 6.0831,
+      "step": 244500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.614305174875675e-06,
+      "loss": 6.0846,
+      "step": 245000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.613484547588177e-06,
+      "loss": 6.0821,
+      "step": 245500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.61266392030068e-06,
+      "loss": 6.0815,
+      "step": 246000
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.61184329301318e-06,
+      "loss": 6.0847,
+      "step": 246500
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.61102266572568e-06,
+      "loss": 6.082,
+      "step": 247000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.610202038438183e-06,
+      "loss": 6.077,
+      "step": 247500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.609381411150685e-06,
+      "loss": 6.0798,
+      "step": 248000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.608560783863185e-06,
+      "loss": 6.0837,
+      "step": 248500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.607740156575688e-06,
+      "loss": 6.0774,
+      "step": 249000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.60691952928819e-06,
+      "loss": 6.0828,
+      "step": 249500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.60609890200069e-06,
+      "loss": 6.0864,
+      "step": 250000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.60527827471319e-06,
+      "loss": 6.08,
+      "step": 250500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.604457647425693e-06,
+      "loss": 6.0765,
+      "step": 251000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.603637020138195e-06,
+      "loss": 6.077,
+      "step": 251500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.602816392850695e-06,
+      "loss": 6.0741,
+      "step": 252000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.601995765563198e-06,
+      "loss": 6.0779,
+      "step": 252500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.6011751382757e-06,
+      "loss": 6.0772,
+      "step": 253000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.6003545109882e-06,
+      "loss": 6.0775,
+      "step": 253500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.5995338837007e-06,
+      "loss": 6.0756,
+      "step": 254000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.598713256413203e-06,
+      "loss": 6.0769,
+      "step": 254500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.597892629125705e-06,
+      "loss": 6.0787,
+      "step": 255000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.597072001838206e-06,
+      "loss": 6.0747,
+      "step": 255500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.596251374550708e-06,
+      "loss": 6.0729,
+      "step": 256000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.59543074726321e-06,
+      "loss": 6.0782,
+      "step": 256500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.59461011997571e-06,
+      "loss": 6.0732,
+      "step": 257000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 9.59378949268821e-06,
+      "loss": 6.0804,
+      "step": 257500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.592968865400713e-06,
+      "loss": 6.0773,
+      "step": 258000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.592148238113215e-06,
+      "loss": 6.0763,
+      "step": 258500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.591327610825716e-06,
+      "loss": 6.0742,
+      "step": 259000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.590506983538218e-06,
+      "loss": 6.0758,
+      "step": 259500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.58968635625072e-06,
+      "loss": 6.0748,
+      "step": 260000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.58886572896322e-06,
+      "loss": 6.0733,
+      "step": 260500
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.588045101675721e-06,
+      "loss": 6.0744,
+      "step": 261000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.587224474388223e-06,
+      "loss": 6.0739,
+      "step": 261500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.586403847100725e-06,
+      "loss": 6.0721,
+      "step": 262000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.585583219813226e-06,
+      "loss": 6.0722,
+      "step": 262500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.584762592525728e-06,
+      "loss": 6.0759,
+      "step": 263000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.58394196523823e-06,
+      "loss": 6.0716,
+      "step": 263500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.58312133795073e-06,
+      "loss": 6.0763,
+      "step": 264000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.582300710663231e-06,
+      "loss": 6.0722,
+      "step": 264500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.581480083375733e-06,
+      "loss": 6.072,
+      "step": 265000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.580659456088235e-06,
+      "loss": 6.0716,
+      "step": 265500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.579838828800736e-06,
+      "loss": 6.0749,
+      "step": 266000
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 9.579018201513238e-06,
+      "loss": 6.0732,
+      "step": 266500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 9.57819757422574e-06,
+      "loss": 6.0717,
+      "step": 267000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.57737694693824e-06,
+      "loss": 6.0706,
+      "step": 267500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 9.576556319650741e-06,
+      "loss": 6.0698,
+      "step": 268000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.575735692363243e-06,
+      "loss": 6.0731,
+      "step": 268500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 9.574915065075745e-06,
+      "loss": 6.0677,
+      "step": 269000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.574094437788246e-06,
+      "loss": 6.076,
+      "step": 269500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.573273810500748e-06,
+      "loss": 6.0716,
+      "step": 270000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.57245318321325e-06,
+      "loss": 6.0679,
+      "step": 270500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.57163255592575e-06,
+      "loss": 6.0697,
+      "step": 271000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.570811928638251e-06,
+      "loss": 6.0709,
+      "step": 271500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 9.569991301350753e-06,
+      "loss": 6.0683,
+      "step": 272000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.569170674063256e-06,
+      "loss": 6.0696,
+      "step": 272500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.568350046775756e-06,
+      "loss": 6.0753,
+      "step": 273000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 9.567529419488258e-06,
+      "loss": 6.0731,
+      "step": 273500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.56670879220076e-06,
+      "loss": 6.0651,
+      "step": 274000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 9.56588816491326e-06,
+      "loss": 6.075,
+      "step": 274500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.565067537625761e-06,
+      "loss": 6.0727,
+      "step": 275000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.564246910338263e-06,
+      "loss": 6.0617,
+      "step": 275500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.563426283050766e-06,
+      "loss": 6.0673,
+      "step": 276000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.562605655763266e-06,
+      "loss": 6.0678,
+      "step": 276500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.561785028475768e-06,
+      "loss": 6.0648,
+      "step": 277000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.56096440118827e-06,
+      "loss": 6.0685,
+      "step": 277500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.560143773900771e-06,
+      "loss": 6.0663,
+      "step": 278000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.559323146613271e-06,
+      "loss": 6.059,
+      "step": 278500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.558502519325774e-06,
+      "loss": 6.0688,
+      "step": 279000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.557681892038276e-06,
+      "loss": 6.0652,
+      "step": 279500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.556861264750776e-06,
+      "loss": 6.0694,
+      "step": 280000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.556040637463278e-06,
+      "loss": 6.0708,
+      "step": 280500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.555220010175779e-06,
+      "loss": 6.0694,
+      "step": 281000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.554399382888281e-06,
+      "loss": 6.0656,
+      "step": 281500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.553578755600781e-06,
+      "loss": 6.0589,
+      "step": 282000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.552758128313284e-06,
+      "loss": 6.0636,
+      "step": 282500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.551937501025786e-06,
+      "loss": 6.0635,
+      "step": 283000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.551116873738286e-06,
+      "loss": 6.0701,
+      "step": 283500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.550296246450788e-06,
+      "loss": 6.0639,
+      "step": 284000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.549475619163289e-06,
+      "loss": 6.0627,
+      "step": 284500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.548654991875791e-06,
+      "loss": 6.0634,
+      "step": 285000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.547834364588292e-06,
+      "loss": 6.0562,
+      "step": 285500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.547013737300794e-06,
+      "loss": 6.0634,
+      "step": 286000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.546193110013296e-06,
+      "loss": 6.0554,
+      "step": 286500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.545372482725796e-06,
+      "loss": 6.0648,
+      "step": 287000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.544551855438298e-06,
+      "loss": 6.0638,
+      "step": 287500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.543731228150799e-06,
+      "loss": 6.0606,
+      "step": 288000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.542910600863301e-06,
+      "loss": 6.0619,
+      "step": 288500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 9.542089973575802e-06,
+      "loss": 6.0612,
+      "step": 289000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 9.541269346288304e-06,
+      "loss": 6.0635,
+      "step": 289500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.540448719000806e-06,
+      "loss": 6.0597,
+      "step": 290000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.539628091713306e-06,
+      "loss": 6.0616,
+      "step": 290500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.538807464425809e-06,
+      "loss": 6.0625,
+      "step": 291000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.537986837138309e-06,
+      "loss": 6.0665,
+      "step": 291500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.537166209850811e-06,
+      "loss": 6.0578,
+      "step": 292000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.536345582563312e-06,
+      "loss": 6.0617,
+      "step": 292500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.535524955275814e-06,
+      "loss": 6.0628,
+      "step": 293000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.534704327988316e-06,
+      "loss": 6.0616,
+      "step": 293500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.533883700700816e-06,
+      "loss": 6.0595,
+      "step": 294000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.533063073413319e-06,
+      "loss": 6.0578,
+      "step": 294500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.532242446125819e-06,
+      "loss": 6.0633,
+      "step": 295000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 9.531421818838321e-06,
+      "loss": 6.0601,
+      "step": 295500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.530601191550822e-06,
+      "loss": 6.0626,
+      "step": 296000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.529780564263324e-06,
+      "loss": 6.0549,
+      "step": 296500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.528959936975824e-06,
+      "loss": 6.055,
+      "step": 297000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.528139309688327e-06,
+      "loss": 6.055,
+      "step": 297500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.527318682400829e-06,
+      "loss": 6.056,
+      "step": 298000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.52649805511333e-06,
+      "loss": 6.0552,
+      "step": 298500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.525677427825831e-06,
+      "loss": 6.0611,
+      "step": 299000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.524856800538332e-06,
+      "loss": 6.0549,
+      "step": 299500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.524036173250834e-06,
+      "loss": 6.0573,
+      "step": 300000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.523215545963334e-06,
+      "loss": 6.0537,
+      "step": 300500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.522394918675837e-06,
+      "loss": 6.0568,
+      "step": 301000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.521574291388339e-06,
+      "loss": 6.0534,
+      "step": 301500
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 9.52075366410084e-06,
+      "loss": 6.0588,
+      "step": 302000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.519933036813341e-06,
+      "loss": 6.0598,
+      "step": 302500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.519112409525842e-06,
+      "loss": 6.0571,
+      "step": 303000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 9.518291782238344e-06,
+      "loss": 6.0576,
+      "step": 303500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.517471154950845e-06,
+      "loss": 6.0549,
+      "step": 304000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.516650527663347e-06,
+      "loss": 6.0574,
+      "step": 304500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.515829900375849e-06,
+      "loss": 6.0587,
+      "step": 305000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.51500927308835e-06,
+      "loss": 6.0511,
+      "step": 305500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.514188645800852e-06,
+      "loss": 6.0537,
+      "step": 306000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.513368018513352e-06,
+      "loss": 6.0539,
+      "step": 306500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.512547391225854e-06,
+      "loss": 6.0567,
+      "step": 307000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.511726763938355e-06,
+      "loss": 6.057,
+      "step": 307500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.510906136650857e-06,
+      "loss": 6.0554,
+      "step": 308000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.510085509363359e-06,
+      "loss": 6.0543,
+      "step": 308500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.50926488207586e-06,
+      "loss": 6.058,
+      "step": 309000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.508444254788362e-06,
+      "loss": 6.0502,
+      "step": 309500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.507623627500862e-06,
+      "loss": 6.0468,
+      "step": 310000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.506803000213364e-06,
+      "loss": 6.0489,
+      "step": 310500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.505982372925865e-06,
+      "loss": 6.0528,
+      "step": 311000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.505161745638367e-06,
+      "loss": 6.0494,
+      "step": 311500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.504341118350869e-06,
+      "loss": 6.056,
+      "step": 312000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.50352049106337e-06,
+      "loss": 6.0498,
+      "step": 312500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.50269986377587e-06,
+      "loss": 6.0537,
+      "step": 313000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.501879236488372e-06,
+      "loss": 6.0493,
+      "step": 313500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.501058609200874e-06,
+      "loss": 6.0481,
+      "step": 314000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.500237981913375e-06,
+      "loss": 6.0534,
+      "step": 314500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.499417354625877e-06,
+      "loss": 6.0497,
+      "step": 315000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.49859672733838e-06,
+      "loss": 6.0498,
+      "step": 315500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.49777610005088e-06,
+      "loss": 6.0508,
+      "step": 316000
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.49695547276338e-06,
+      "loss": 6.0472,
+      "step": 316500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.496134845475882e-06,
+      "loss": 6.0531,
+      "step": 317000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.495314218188384e-06,
+      "loss": 6.0525,
+      "step": 317500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.494493590900885e-06,
+      "loss": 6.054,
+      "step": 318000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.493672963613387e-06,
+      "loss": 6.0475,
+      "step": 318500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.49285233632589e-06,
+      "loss": 6.0485,
+      "step": 319000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.49203170903839e-06,
+      "loss": 6.044,
+      "step": 319500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.49121108175089e-06,
+      "loss": 6.0459,
+      "step": 320000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.490390454463392e-06,
+      "loss": 6.0456,
+      "step": 320500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.489569827175895e-06,
+      "loss": 6.0437,
+      "step": 321000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.488749199888395e-06,
+      "loss": 6.051,
+      "step": 321500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.487928572600897e-06,
+      "loss": 6.0495,
+      "step": 322000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.4871079453134e-06,
+      "loss": 6.0497,
+      "step": 322500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.4862873180259e-06,
+      "loss": 6.0444,
+      "step": 323000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.4854666907384e-06,
+      "loss": 6.0482,
+      "step": 323500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.484646063450902e-06,
+      "loss": 6.0472,
+      "step": 324000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.483825436163405e-06,
+      "loss": 6.0416,
+      "step": 324500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.483004808875905e-06,
+      "loss": 6.0487,
+      "step": 325000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.482184181588407e-06,
+      "loss": 6.0468,
+      "step": 325500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.48136355430091e-06,
+      "loss": 6.0461,
+      "step": 326000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.48054292701341e-06,
+      "loss": 6.0487,
+      "step": 326500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.47972229972591e-06,
+      "loss": 6.0504,
+      "step": 327000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.478901672438413e-06,
+      "loss": 6.0414,
+      "step": 327500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.478081045150915e-06,
+      "loss": 6.0444,
+      "step": 328000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.477260417863415e-06,
+      "loss": 6.0426,
+      "step": 328500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.476439790575916e-06,
+      "loss": 6.0453,
+      "step": 329000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.47561916328842e-06,
+      "loss": 6.0413,
+      "step": 329500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.47479853600092e-06,
+      "loss": 6.0487,
+      "step": 330000
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.47397790871342e-06,
+      "loss": 6.0438,
+      "step": 330500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.473157281425923e-06,
+      "loss": 6.0405,
+      "step": 331000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.472336654138425e-06,
+      "loss": 6.0385,
+      "step": 331500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.471516026850925e-06,
+      "loss": 6.0419,
+      "step": 332000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.470695399563426e-06,
+      "loss": 6.0424,
+      "step": 332500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.46987477227593e-06,
+      "loss": 6.0349,
+      "step": 333000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.46905414498843e-06,
+      "loss": 6.042,
+      "step": 333500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.46823351770093e-06,
+      "loss": 6.041,
+      "step": 334000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.467412890413433e-06,
+      "loss": 6.0417,
+      "step": 334500
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.466592263125935e-06,
+      "loss": 6.0393,
+      "step": 335000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.465771635838435e-06,
+      "loss": 6.0396,
+      "step": 335500
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 9.464951008550936e-06,
+      "loss": 6.041,
+      "step": 336000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 9.46413038126344e-06,
+      "loss": 6.0361,
+      "step": 336500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 9.46330975397594e-06,
+      "loss": 6.0383,
+      "step": 337000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 9.46248912668844e-06,
+      "loss": 6.0385,
+      "step": 337500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 9.461668499400943e-06,
+      "loss": 6.041,
+      "step": 338000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 9.460847872113445e-06,
+      "loss": 6.0398,
+      "step": 338500
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 9.460027244825945e-06,
+      "loss": 6.0345,
+      "step": 339000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 9.459206617538446e-06,
+      "loss": 6.0355,
+      "step": 339500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 9.45838599025095e-06,
+      "loss": 6.0397,
+      "step": 340000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 9.45756536296345e-06,
+      "loss": 6.0371,
+      "step": 340500
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 9.45674473567595e-06,
+      "loss": 6.0376,
+      "step": 341000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 9.455924108388453e-06,
+      "loss": 6.0431,
+      "step": 341500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 9.455103481100955e-06,
+      "loss": 6.038,
+      "step": 342000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 9.454282853813456e-06,
+      "loss": 6.0362,
+      "step": 342500
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 9.453462226525956e-06,
+      "loss": 6.0377,
+      "step": 343000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 9.45264159923846e-06,
+      "loss": 6.0291,
+      "step": 343500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 9.45182097195096e-06,
+      "loss": 6.0321,
+      "step": 344000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 9.45100034466346e-06,
+      "loss": 6.0318,
+      "step": 344500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 9.450179717375963e-06,
+      "loss": 6.0313,
+      "step": 345000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 9.449359090088465e-06,
+      "loss": 6.0327,
+      "step": 345500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 9.448538462800966e-06,
+      "loss": 6.0329,
+      "step": 346000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 9.447717835513466e-06,
+      "loss": 6.0313,
+      "step": 346500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 9.44689720822597e-06,
+      "loss": 6.0282,
+      "step": 347000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 9.44607658093847e-06,
+      "loss": 6.0267,
+      "step": 347500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 9.445255953650971e-06,
+      "loss": 6.0343,
+      "step": 348000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 9.444435326363473e-06,
+      "loss": 6.032,
+      "step": 348500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 9.443614699075975e-06,
+      "loss": 6.0323,
+      "step": 349000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.442794071788476e-06,
+      "loss": 6.0306,
+      "step": 349500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.441973444500976e-06,
+      "loss": 6.0243,
+      "step": 350000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.44115281721348e-06,
+      "loss": 6.0288,
+      "step": 350500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.44033218992598e-06,
+      "loss": 6.0224,
+      "step": 351000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 9.439511562638481e-06,
+      "loss": 6.0283,
+      "step": 351500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 9.438690935350983e-06,
+      "loss": 6.0301,
+      "step": 352000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.437870308063485e-06,
+      "loss": 6.0284,
+      "step": 352500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.437049680775986e-06,
+      "loss": 6.0203,
+      "step": 353000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 9.436229053488486e-06,
+      "loss": 6.023,
+      "step": 353500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.43540842620099e-06,
+      "loss": 6.0231,
+      "step": 354000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.43458779891349e-06,
+      "loss": 6.0216,
+      "step": 354500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.433767171625991e-06,
+      "loss": 6.0246,
+      "step": 355000
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.432946544338493e-06,
+      "loss": 6.0218,
+      "step": 355500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.432125917050995e-06,
+      "loss": 6.0241,
+      "step": 356000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 9.431305289763496e-06,
+      "loss": 6.0213,
+      "step": 356500
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 9.430484662475996e-06,
+      "loss": 6.0216,
+      "step": 357000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 9.4296640351885e-06,
+      "loss": 6.0221,
+      "step": 357500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 9.428843407901e-06,
+      "loss": 6.0203,
+      "step": 358000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 9.428022780613501e-06,
+      "loss": 6.0204,
+      "step": 358500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 9.427202153326003e-06,
+      "loss": 6.0172,
+      "step": 359000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.426381526038505e-06,
+      "loss": 6.0179,
+      "step": 359500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 9.425560898751006e-06,
+      "loss": 6.0171,
+      "step": 360000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.424740271463506e-06,
+      "loss": 6.021,
+      "step": 360500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.42391964417601e-06,
+      "loss": 6.0185,
+      "step": 361000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.42309901688851e-06,
+      "loss": 6.0217,
+      "step": 361500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 9.422278389601011e-06,
+      "loss": 6.0146,
+      "step": 362000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 9.421457762313513e-06,
+      "loss": 6.0137,
+      "step": 362500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 9.420637135026016e-06,
+      "loss": 6.0168,
+      "step": 363000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 9.419816507738516e-06,
+      "loss": 6.015,
+      "step": 363500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 9.418995880451016e-06,
+      "loss": 6.0169,
+      "step": 364000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.418175253163519e-06,
+      "loss": 6.0149,
+      "step": 364500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 9.41735462587602e-06,
+      "loss": 6.016,
+      "step": 365000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.416533998588521e-06,
+      "loss": 6.0161,
+      "step": 365500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 9.415713371301023e-06,
+      "loss": 6.0164,
+      "step": 366000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 9.414892744013526e-06,
+      "loss": 6.0162,
+      "step": 366500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 9.414072116726026e-06,
+      "loss": 6.0139,
+      "step": 367000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 9.413251489438527e-06,
+      "loss": 6.0092,
+      "step": 367500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 9.412430862151029e-06,
+      "loss": 6.0145,
+      "step": 368000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 9.411610234863531e-06,
+      "loss": 6.012,
+      "step": 368500
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 9.410789607576031e-06,
+      "loss": 6.012,
+      "step": 369000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 9.409968980288534e-06,
+      "loss": 6.0128,
+      "step": 369500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 9.409148353001036e-06,
+      "loss": 6.0143,
+      "step": 370000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 9.408327725713536e-06,
+      "loss": 6.0091,
+      "step": 370500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 9.407507098426037e-06,
+      "loss": 6.014,
+      "step": 371000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 9.406686471138539e-06,
+      "loss": 6.006,
+      "step": 371500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.405865843851041e-06,
+      "loss": 6.0097,
+      "step": 372000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.405045216563541e-06,
+      "loss": 6.0092,
+      "step": 372500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 9.404224589276044e-06,
+      "loss": 6.0093,
+      "step": 373000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.403403961988546e-06,
+      "loss": 6.0086,
+      "step": 373500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 9.402583334701046e-06,
+      "loss": 6.0118,
+      "step": 374000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.401762707413547e-06,
+      "loss": 6.0096,
+      "step": 374500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.400942080126049e-06,
+      "loss": 6.0038,
+      "step": 375000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.400121452838551e-06,
+      "loss": 6.0067,
+      "step": 375500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.399300825551052e-06,
+      "loss": 6.0068,
+      "step": 376000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.398480198263554e-06,
+      "loss": 6.0066,
+      "step": 376500
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 9.397659570976056e-06,
+      "loss": 6.0053,
+      "step": 377000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.396838943688556e-06,
+      "loss": 6.0094,
+      "step": 377500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.396018316401057e-06,
+      "loss": 6.0098,
+      "step": 378000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 9.395197689113559e-06,
+      "loss": 6.007,
+      "step": 378500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 9.394377061826061e-06,
+      "loss": 6.0012,
+      "step": 379000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.393556434538562e-06,
+      "loss": 6.0083,
+      "step": 379500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.392735807251064e-06,
+      "loss": 6.0067,
+      "step": 380000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.391915179963564e-06,
+      "loss": 6.0015,
+      "step": 380500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 9.391094552676066e-06,
+      "loss": 6.0052,
+      "step": 381000
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 9.390273925388567e-06,
+      "loss": 6.0016,
+      "step": 381500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 9.389453298101069e-06,
+      "loss": 6.0042,
+      "step": 382000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 9.388632670813571e-06,
+      "loss": 6.0034,
+      "step": 382500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 9.387812043526072e-06,
+      "loss": 6.0036,
+      "step": 383000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 9.386991416238574e-06,
+      "loss": 6.0028,
+      "step": 383500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.386170788951074e-06,
+      "loss": 6.0006,
+      "step": 384000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.385350161663577e-06,
+      "loss": 6.0023,
+      "step": 384500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 9.384529534376077e-06,
+      "loss": 6.0028,
+      "step": 385000
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 9.38370890708858e-06,
+      "loss": 6.0038,
+      "step": 385500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 9.382888279801081e-06,
+      "loss": 5.9974,
+      "step": 386000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 9.382067652513582e-06,
+      "loss": 6.0004,
+      "step": 386500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 9.381247025226084e-06,
+      "loss": 5.9975,
+      "step": 387000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 9.380426397938584e-06,
+      "loss": 6.0015,
+      "step": 387500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 9.379605770651087e-06,
+      "loss": 6.0018,
+      "step": 388000
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 9.378785143363587e-06,
+      "loss": 6.0008,
+      "step": 388500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 9.37796451607609e-06,
+      "loss": 5.9974,
+      "step": 389000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 9.377143888788591e-06,
+      "loss": 5.9938,
+      "step": 389500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 9.376323261501092e-06,
+      "loss": 5.9963,
+      "step": 390000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 9.375502634213594e-06,
+      "loss": 5.9935,
+      "step": 390500
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 9.374682006926095e-06,
+      "loss": 5.9996,
+      "step": 391000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 9.373861379638597e-06,
+      "loss": 5.994,
+      "step": 391500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 9.373040752351097e-06,
+      "loss": 5.9927,
+      "step": 392000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 9.3722201250636e-06,
+      "loss": 5.9924,
+      "step": 392500
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 9.371399497776102e-06,
+      "loss": 5.9908,
+      "step": 393000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 9.370578870488602e-06,
+      "loss": 5.992,
+      "step": 393500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 9.369758243201104e-06,
+      "loss": 5.9866,
+      "step": 394000
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 9.368937615913605e-06,
+      "loss": 5.9882,
+      "step": 394500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 9.368116988626107e-06,
+      "loss": 5.9891,
+      "step": 395000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 9.367296361338607e-06,
+      "loss": 5.9851,
+      "step": 395500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 9.36647573405111e-06,
+      "loss": 5.9838,
+      "step": 396000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 9.36565510676361e-06,
+      "loss": 5.9886,
+      "step": 396500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 9.364834479476112e-06,
+      "loss": 5.9866,
+      "step": 397000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 9.364013852188614e-06,
+      "loss": 5.9867,
+      "step": 397500
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 9.363193224901115e-06,
+      "loss": 5.9808,
+      "step": 398000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 9.362372597613617e-06,
+      "loss": 5.9816,
+      "step": 398500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 9.361551970326117e-06,
+      "loss": 5.984,
+      "step": 399000
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 9.36073134303862e-06,
+      "loss": 5.9744,
+      "step": 399500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 9.35991071575112e-06,
+      "loss": 5.9796,
+      "step": 400000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 9.359090088463622e-06,
+      "loss": 5.977,
+      "step": 400500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 9.358269461176124e-06,
+      "loss": 5.9781,
+      "step": 401000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 9.357448833888625e-06,
+      "loss": 5.9776,
+      "step": 401500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 9.356628206601127e-06,
+      "loss": 5.9737,
+      "step": 402000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 9.355807579313627e-06,
+      "loss": 5.9788,
+      "step": 402500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 9.35498695202613e-06,
+      "loss": 5.9798,
+      "step": 403000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 9.35416632473863e-06,
+      "loss": 5.9724,
+      "step": 403500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 9.353345697451132e-06,
+      "loss": 5.9726,
+      "step": 404000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 9.352525070163634e-06,
+      "loss": 5.975,
+      "step": 404500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 9.351704442876135e-06,
+      "loss": 5.9721,
+      "step": 405000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 9.350883815588637e-06,
+      "loss": 5.9684,
+      "step": 405500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 9.350063188301138e-06,
+      "loss": 5.9682,
+      "step": 406000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 9.34924256101364e-06,
+      "loss": 5.9742,
+      "step": 406500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 9.34842193372614e-06,
+      "loss": 5.9627,
+      "step": 407000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.347601306438642e-06,
+      "loss": 5.9704,
+      "step": 407500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.346780679151144e-06,
+      "loss": 5.9664,
+      "step": 408000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.345960051863645e-06,
+      "loss": 5.968,
+      "step": 408500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.345139424576147e-06,
+      "loss": 5.9674,
+      "step": 409000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.344318797288648e-06,
+      "loss": 5.9636,
+      "step": 409500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.34349817000115e-06,
+      "loss": 5.9667,
+      "step": 410000
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.34267754271365e-06,
+      "loss": 5.9664,
+      "step": 410500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.341856915426152e-06,
+      "loss": 5.9651,
+      "step": 411000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.341036288138655e-06,
+      "loss": 5.9607,
+      "step": 411500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.340215660851155e-06,
+      "loss": 5.9648,
+      "step": 412000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.339395033563656e-06,
+      "loss": 5.9607,
+      "step": 412500
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.338574406276158e-06,
+      "loss": 5.9614,
+      "step": 413000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.33775377898866e-06,
+      "loss": 5.9642,
+      "step": 413500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.33693315170116e-06,
+      "loss": 5.9609,
+      "step": 414000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.336112524413662e-06,
+      "loss": 5.9622,
+      "step": 414500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.335291897126165e-06,
+      "loss": 5.9565,
+      "step": 415000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.334471269838665e-06,
+      "loss": 5.9607,
+      "step": 415500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.333650642551166e-06,
+      "loss": 5.9566,
+      "step": 416000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.332830015263668e-06,
+      "loss": 5.9552,
+      "step": 416500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.33200938797617e-06,
+      "loss": 5.9557,
+      "step": 417000
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.33118876068867e-06,
+      "loss": 5.957,
+      "step": 417500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.330368133401173e-06,
+      "loss": 5.9545,
+      "step": 418000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.329547506113675e-06,
+      "loss": 5.9543,
+      "step": 418500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.328726878826175e-06,
+      "loss": 5.9561,
+      "step": 419000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.327906251538676e-06,
+      "loss": 5.9581,
+      "step": 419500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.327085624251178e-06,
+      "loss": 5.9493,
+      "step": 420000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.32626499696368e-06,
+      "loss": 5.9466,
+      "step": 420500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.32544436967618e-06,
+      "loss": 5.9471,
+      "step": 421000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.324623742388683e-06,
+      "loss": 5.9528,
+      "step": 421500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.323803115101185e-06,
+      "loss": 5.9487,
+      "step": 422000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.322982487813685e-06,
+      "loss": 5.9496,
+      "step": 422500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.322161860526186e-06,
+      "loss": 5.9479,
+      "step": 423000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.321341233238688e-06,
+      "loss": 5.9474,
+      "step": 423500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.32052060595119e-06,
+      "loss": 5.9511,
+      "step": 424000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.31969997866369e-06,
+      "loss": 5.9464,
+      "step": 424500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.318879351376193e-06,
+      "loss": 5.9462,
+      "step": 425000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.318058724088695e-06,
+      "loss": 5.9464,
+      "step": 425500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.317238096801195e-06,
+      "loss": 5.9428,
+      "step": 426000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.316417469513696e-06,
+      "loss": 5.9441,
+      "step": 426500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.315596842226198e-06,
+      "loss": 5.9412,
+      "step": 427000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.3147762149387e-06,
+      "loss": 5.9389,
+      "step": 427500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.3139555876512e-06,
+      "loss": 5.9413,
+      "step": 428000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.313134960363703e-06,
+      "loss": 5.9364,
+      "step": 428500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.312314333076205e-06,
+      "loss": 5.9398,
+      "step": 429000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.311493705788705e-06,
+      "loss": 5.9362,
+      "step": 429500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.310673078501206e-06,
+      "loss": 5.9414,
+      "step": 430000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.30985245121371e-06,
+      "loss": 5.9381,
+      "step": 430500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.30903182392621e-06,
+      "loss": 5.9348,
+      "step": 431000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.30821119663871e-06,
+      "loss": 5.9402,
+      "step": 431500
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.307390569351213e-06,
+      "loss": 5.9379,
+      "step": 432000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.306569942063715e-06,
+      "loss": 5.933,
+      "step": 432500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.305749314776216e-06,
+      "loss": 5.9344,
+      "step": 433000
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.304928687488716e-06,
+      "loss": 5.9338,
+      "step": 433500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.30410806020122e-06,
+      "loss": 5.9278,
+      "step": 434000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.30328743291372e-06,
+      "loss": 5.9389,
+      "step": 434500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.30246680562622e-06,
+      "loss": 5.9304,
+      "step": 435000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.301646178338723e-06,
+      "loss": 5.9326,
+      "step": 435500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.300825551051225e-06,
+      "loss": 5.9289,
+      "step": 436000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.300004923763726e-06,
+      "loss": 5.9338,
+      "step": 436500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.299184296476226e-06,
+      "loss": 5.9316,
+      "step": 437000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.29836366918873e-06,
+      "loss": 5.9276,
+      "step": 437500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.29754304190123e-06,
+      "loss": 5.9242,
+      "step": 438000
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.296722414613731e-06,
+      "loss": 5.9324,
+      "step": 438500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.295901787326233e-06,
+      "loss": 5.9282,
+      "step": 439000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.295081160038735e-06,
+      "loss": 5.925,
+      "step": 439500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.294260532751236e-06,
+      "loss": 5.922,
+      "step": 440000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.293439905463736e-06,
+      "loss": 5.9273,
+      "step": 440500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.29261927817624e-06,
+      "loss": 5.9262,
+      "step": 441000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.29179865088874e-06,
+      "loss": 5.9157,
+      "step": 441500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.290978023601241e-06,
+      "loss": 5.9213,
+      "step": 442000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.290157396313743e-06,
+      "loss": 5.9233,
+      "step": 442500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.289336769026245e-06,
+      "loss": 5.9215,
+      "step": 443000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.288516141738746e-06,
+      "loss": 5.9176,
+      "step": 443500
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.287695514451246e-06,
+      "loss": 5.9196,
+      "step": 444000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.286874887163748e-06,
+      "loss": 5.9146,
+      "step": 444500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.28605425987625e-06,
+      "loss": 5.9157,
+      "step": 445000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.285233632588751e-06,
+      "loss": 5.9167,
+      "step": 445500
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 9.284413005301253e-06,
+      "loss": 5.9171,
+      "step": 446000
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 9.283592378013755e-06,
+      "loss": 5.913,
+      "step": 446500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 9.282771750726256e-06,
+      "loss": 5.9099,
+      "step": 447000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 9.281951123438756e-06,
+      "loss": 5.9192,
+      "step": 447500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 9.281130496151259e-06,
+      "loss": 5.9136,
+      "step": 448000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 9.28030986886376e-06,
+      "loss": 5.9143,
+      "step": 448500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 9.279489241576261e-06,
+      "loss": 5.9138,
+      "step": 449000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 9.278668614288763e-06,
+      "loss": 5.9083,
+      "step": 449500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 9.277847987001266e-06,
+      "loss": 5.9124,
+      "step": 450000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 9.277027359713766e-06,
+      "loss": 5.9162,
+      "step": 450500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 9.276206732426266e-06,
+      "loss": 5.9092,
+      "step": 451000
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 9.275386105138769e-06,
+      "loss": 5.906,
+      "step": 451500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 9.27456547785127e-06,
+      "loss": 5.9078,
+      "step": 452000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 9.273744850563771e-06,
+      "loss": 5.9039,
+      "step": 452500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 9.272924223276273e-06,
+      "loss": 5.9066,
+      "step": 453000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 9.272103595988776e-06,
+      "loss": 5.9042,
+      "step": 453500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 9.271282968701276e-06,
+      "loss": 5.9054,
+      "step": 454000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 9.270462341413777e-06,
+      "loss": 5.9078,
+      "step": 454500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 9.269641714126279e-06,
+      "loss": 5.9045,
+      "step": 455000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 9.268821086838781e-06,
+      "loss": 5.9066,
+      "step": 455500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 9.268000459551281e-06,
+      "loss": 5.904,
+      "step": 456000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 9.267179832263784e-06,
+      "loss": 5.9075,
+      "step": 456500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 9.266359204976286e-06,
+      "loss": 5.9052,
+      "step": 457000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 9.265538577688786e-06,
+      "loss": 5.9058,
+      "step": 457500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 9.264717950401287e-06,
+      "loss": 5.902,
+      "step": 458000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 9.263897323113789e-06,
+      "loss": 5.9002,
+      "step": 458500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 9.263076695826291e-06,
+      "loss": 5.9047,
+      "step": 459000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 9.262256068538791e-06,
+      "loss": 5.8994,
+      "step": 459500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 9.261435441251294e-06,
+      "loss": 5.9009,
+      "step": 460000
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 9.260614813963794e-06,
+      "loss": 5.8997,
+      "step": 460500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 9.259794186676296e-06,
+      "loss": 5.9011,
+      "step": 461000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 9.258973559388797e-06,
+      "loss": 5.9017,
+      "step": 461500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 9.258152932101299e-06,
+      "loss": 5.8942,
+      "step": 462000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 9.257332304813801e-06,
+      "loss": 5.891,
+      "step": 462500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 9.256511677526302e-06,
+      "loss": 5.8959,
+      "step": 463000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 9.255691050238804e-06,
+      "loss": 5.8971,
+      "step": 463500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 9.254870422951304e-06,
+      "loss": 5.8956,
+      "step": 464000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 9.254049795663806e-06,
+      "loss": 5.8963,
+      "step": 464500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 9.253229168376307e-06,
+      "loss": 5.8939,
+      "step": 465000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 9.252408541088809e-06,
+      "loss": 5.8924,
+      "step": 465500
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 9.251587913801311e-06,
+      "loss": 5.8925,
+      "step": 466000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 9.250767286513812e-06,
+      "loss": 5.8917,
+      "step": 466500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 9.249946659226314e-06,
+      "loss": 5.8877,
+      "step": 467000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 9.249126031938814e-06,
+      "loss": 5.8885,
+      "step": 467500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 9.248305404651316e-06,
+      "loss": 5.889,
+      "step": 468000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 9.247484777363817e-06,
+      "loss": 5.8897,
+      "step": 468500
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 9.246664150076319e-06,
+      "loss": 5.8855,
+      "step": 469000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 9.245843522788821e-06,
+      "loss": 5.8875,
+      "step": 469500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 9.245022895501322e-06,
+      "loss": 5.8856,
+      "step": 470000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 9.244202268213824e-06,
+      "loss": 5.8898,
+      "step": 470500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 9.243381640926324e-06,
+      "loss": 5.8835,
+      "step": 471000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.242561013638826e-06,
+      "loss": 5.8842,
+      "step": 471500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.241740386351327e-06,
+      "loss": 5.8814,
+      "step": 472000
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.240919759063829e-06,
+      "loss": 5.8826,
+      "step": 472500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.240099131776331e-06,
+      "loss": 5.8823,
+      "step": 473000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.239278504488832e-06,
+      "loss": 5.882,
+      "step": 473500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.238457877201334e-06,
+      "loss": 5.8801,
+      "step": 474000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.237637249913834e-06,
+      "loss": 5.8825,
+      "step": 474500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.236816622626337e-06,
+      "loss": 5.8797,
+      "step": 475000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.235995995338837e-06,
+      "loss": 5.8797,
+      "step": 475500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.23517536805134e-06,
+      "loss": 5.8774,
+      "step": 476000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.23435474076384e-06,
+      "loss": 5.8806,
+      "step": 476500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.233534113476342e-06,
+      "loss": 5.8793,
+      "step": 477000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.232713486188844e-06,
+      "loss": 5.8801,
+      "step": 477500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.231892858901344e-06,
+      "loss": 5.8765,
+      "step": 478000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.231072231613847e-06,
+      "loss": 5.8781,
+      "step": 478500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.230251604326347e-06,
+      "loss": 5.8772,
+      "step": 479000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.22943097703885e-06,
+      "loss": 5.8784,
+      "step": 479500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.22861034975135e-06,
+      "loss": 5.8737,
+      "step": 480000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.227789722463852e-06,
+      "loss": 5.8762,
+      "step": 480500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.226969095176354e-06,
+      "loss": 5.8727,
+      "step": 481000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.226148467888855e-06,
+      "loss": 5.8727,
+      "step": 481500
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.225327840601357e-06,
+      "loss": 5.8751,
+      "step": 482000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.224507213313857e-06,
+      "loss": 5.8718,
+      "step": 482500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.22368658602636e-06,
+      "loss": 5.8712,
+      "step": 483000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.22286595873886e-06,
+      "loss": 5.8723,
+      "step": 483500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.222045331451362e-06,
+      "loss": 5.8761,
+      "step": 484000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.221224704163864e-06,
+      "loss": 5.8681,
+      "step": 484500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.220404076876365e-06,
+      "loss": 5.8711,
+      "step": 485000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.219583449588867e-06,
+      "loss": 5.8706,
+      "step": 485500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.218762822301367e-06,
+      "loss": 5.8665,
+      "step": 486000
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.21794219501387e-06,
+      "loss": 5.8697,
+      "step": 486500
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.21712156772637e-06,
+      "loss": 5.8658,
+      "step": 487000
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.216300940438872e-06,
+      "loss": 5.8676,
+      "step": 487500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.215480313151374e-06,
+      "loss": 5.8692,
+      "step": 488000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.214659685863875e-06,
+      "loss": 5.8653,
+      "step": 488500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.213839058576377e-06,
+      "loss": 5.8681,
+      "step": 489000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.213018431288877e-06,
+      "loss": 5.8624,
+      "step": 489500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.21219780400138e-06,
+      "loss": 5.8675,
+      "step": 490000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.21137717671388e-06,
+      "loss": 5.8598,
+      "step": 490500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.210556549426382e-06,
+      "loss": 5.8649,
+      "step": 491000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.209735922138884e-06,
+      "loss": 5.8649,
+      "step": 491500
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.208915294851385e-06,
+      "loss": 5.8613,
+      "step": 492000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.208094667563885e-06,
+      "loss": 5.8608,
+      "step": 492500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.207274040276387e-06,
+      "loss": 5.8613,
+      "step": 493000
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.20645341298889e-06,
+      "loss": 5.8586,
+      "step": 493500
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.20563278570139e-06,
+      "loss": 5.8593,
+      "step": 494000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.204812158413892e-06,
+      "loss": 5.8597,
+      "step": 494500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.203991531126394e-06,
+      "loss": 5.8598,
+      "step": 495000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.203170903838895e-06,
+      "loss": 5.8576,
+      "step": 495500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.202350276551395e-06,
+      "loss": 5.8591,
+      "step": 496000
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.201529649263898e-06,
+      "loss": 5.8588,
+      "step": 496500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.2007090219764e-06,
+      "loss": 5.8583,
+      "step": 497000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.1998883946889e-06,
+      "loss": 5.8579,
+      "step": 497500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.199067767401402e-06,
+      "loss": 5.858,
+      "step": 498000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.198247140113905e-06,
+      "loss": 5.8586,
+      "step": 498500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.197426512826405e-06,
+      "loss": 5.856,
+      "step": 499000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.196605885538905e-06,
+      "loss": 5.8592,
+      "step": 499500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.195785258251408e-06,
+      "loss": 5.8532,
+      "step": 500000
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.19496463096391e-06,
+      "loss": 5.8557,
+      "step": 500500
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.19414400367641e-06,
+      "loss": 5.8584,
+      "step": 501000
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.193323376388912e-06,
+      "loss": 5.8541,
+      "step": 501500
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 9.192502749101415e-06,
+      "loss": 5.853,
+      "step": 502000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 9.191682121813915e-06,
+      "loss": 5.8516,
+      "step": 502500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 9.190861494526416e-06,
+      "loss": 5.8536,
+      "step": 503000
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 9.190040867238918e-06,
+      "loss": 5.8525,
+      "step": 503500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 9.18922023995142e-06,
+      "loss": 5.8535,
+      "step": 504000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 9.18839961266392e-06,
+      "loss": 5.8511,
+      "step": 504500
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 9.187578985376423e-06,
+      "loss": 5.8508,
+      "step": 505000
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 9.186758358088925e-06,
+      "loss": 5.8532,
+      "step": 505500
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 9.185937730801425e-06,
+      "loss": 5.8523,
+      "step": 506000
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 9.185117103513926e-06,
+      "loss": 5.8531,
+      "step": 506500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 9.184296476226428e-06,
+      "loss": 5.8518,
+      "step": 507000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 9.18347584893893e-06,
+      "loss": 5.8496,
+      "step": 507500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 9.18265522165143e-06,
+      "loss": 5.8476,
+      "step": 508000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 9.181834594363933e-06,
+      "loss": 5.8473,
+      "step": 508500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 9.181013967076435e-06,
+      "loss": 5.8438,
+      "step": 509000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 9.180193339788935e-06,
+      "loss": 5.849,
+      "step": 509500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 9.179372712501436e-06,
+      "loss": 5.8499,
+      "step": 510000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 9.178552085213938e-06,
+      "loss": 5.8455,
+      "step": 510500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 9.17773145792644e-06,
+      "loss": 5.8459,
+      "step": 511000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 9.17691083063894e-06,
+      "loss": 5.8436,
+      "step": 511500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 9.176090203351443e-06,
+      "loss": 5.8442,
+      "step": 512000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 9.175269576063945e-06,
+      "loss": 5.8423,
+      "step": 512500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 9.174448948776445e-06,
+      "loss": 5.8445,
+      "step": 513000
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 9.173628321488946e-06,
+      "loss": 5.8395,
+      "step": 513500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 9.172807694201448e-06,
+      "loss": 5.844,
+      "step": 514000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 9.17198706691395e-06,
+      "loss": 5.8414,
+      "step": 514500
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 9.17116643962645e-06,
+      "loss": 5.8351,
+      "step": 515000
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 9.170345812338953e-06,
+      "loss": 5.8373,
+      "step": 515500
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 9.169525185051455e-06,
+      "loss": 5.8446,
+      "step": 516000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 9.168704557763955e-06,
+      "loss": 5.8411,
+      "step": 516500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 9.167883930476456e-06,
+      "loss": 5.8396,
+      "step": 517000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 9.167063303188958e-06,
+      "loss": 5.8423,
+      "step": 517500
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 9.16624267590146e-06,
+      "loss": 5.8386,
+      "step": 518000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 9.16542204861396e-06,
+      "loss": 5.8388,
+      "step": 518500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 9.164601421326463e-06,
+      "loss": 5.8411,
+      "step": 519000
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 9.163780794038965e-06,
+      "loss": 5.8369,
+      "step": 519500
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 9.162960166751466e-06,
+      "loss": 5.838,
+      "step": 520000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 9.162139539463966e-06,
+      "loss": 5.8323,
+      "step": 520500
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 9.161318912176468e-06,
+      "loss": 5.8284,
+      "step": 521000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 9.16049828488897e-06,
+      "loss": 5.8373,
+      "step": 521500
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 9.15967765760147e-06,
+      "loss": 5.8368,
+      "step": 522000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 9.158857030313973e-06,
+      "loss": 5.8352,
+      "step": 522500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 9.158036403026475e-06,
+      "loss": 5.8339,
+      "step": 523000
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 9.157215775738976e-06,
+      "loss": 5.8336,
+      "step": 523500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.156395148451476e-06,
+      "loss": 5.834,
+      "step": 524000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 9.155574521163978e-06,
+      "loss": 5.8338,
+      "step": 524500
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 9.15475389387648e-06,
+      "loss": 5.8293,
+      "step": 525000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 9.153933266588981e-06,
+      "loss": 5.8311,
+      "step": 525500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.153112639301483e-06,
+      "loss": 5.8325,
+      "step": 526000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.152292012013985e-06,
+      "loss": 5.8319,
+      "step": 526500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.151471384726486e-06,
+      "loss": 5.8352,
+      "step": 527000
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.150650757438986e-06,
+      "loss": 5.8296,
+      "step": 527500
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 9.149830130151488e-06,
+      "loss": 5.83,
+      "step": 528000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.14900950286399e-06,
+      "loss": 5.8285,
+      "step": 528500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 9.148188875576491e-06,
+      "loss": 5.8306,
+      "step": 529000
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 9.147368248288993e-06,
+      "loss": 5.8297,
+      "step": 529500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 9.146547621001495e-06,
+      "loss": 5.8316,
+      "step": 530000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 9.145726993713996e-06,
+      "loss": 5.8238,
+      "step": 530500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 9.144906366426496e-06,
+      "loss": 5.8317,
+      "step": 531000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 9.144085739138998e-06,
+      "loss": 5.8262,
+      "step": 531500
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 9.1432651118515e-06,
+      "loss": 5.8241,
+      "step": 532000
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 9.142444484564001e-06,
+      "loss": 5.827,
+      "step": 532500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 9.141623857276503e-06,
+      "loss": 5.8252,
+      "step": 533000
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 9.140803229989005e-06,
+      "loss": 5.8236,
+      "step": 533500
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 9.139982602701506e-06,
+      "loss": 5.8277,
+      "step": 534000
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 9.139161975414006e-06,
+      "loss": 5.825,
+      "step": 534500
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 9.138341348126508e-06,
+      "loss": 5.8207,
+      "step": 535000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 9.13752072083901e-06,
+      "loss": 5.8177,
+      "step": 535500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 9.136700093551511e-06,
+      "loss": 5.8248,
+      "step": 536000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 9.135879466264013e-06,
+      "loss": 5.8203,
+      "step": 536500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 9.135058838976515e-06,
+      "loss": 5.8238,
+      "step": 537000
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 9.134238211689016e-06,
+      "loss": 5.8202,
+      "step": 537500
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 9.133417584401516e-06,
+      "loss": 5.8215,
+      "step": 538000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 9.132596957114019e-06,
+      "loss": 5.82,
+      "step": 538500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 9.13177632982652e-06,
+      "loss": 5.8205,
+      "step": 539000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 9.130955702539021e-06,
+      "loss": 5.82,
+      "step": 539500
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 9.130135075251523e-06,
+      "loss": 5.8223,
+      "step": 540000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.129314447964026e-06,
+      "loss": 5.8148,
+      "step": 540500
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 9.128493820676526e-06,
+      "loss": 5.8182,
+      "step": 541000
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 9.127673193389026e-06,
+      "loss": 5.8144,
+      "step": 541500
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 9.126852566101529e-06,
+      "loss": 5.8213,
+      "step": 542000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 9.12603193881403e-06,
+      "loss": 5.8135,
+      "step": 542500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 9.125211311526531e-06,
+      "loss": 5.8179,
+      "step": 543000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 9.124390684239033e-06,
+      "loss": 5.8172,
+      "step": 543500
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 9.123570056951534e-06,
+      "loss": 5.8126,
+      "step": 544000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 9.122749429664036e-06,
+      "loss": 5.8131,
+      "step": 544500
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 9.121928802376537e-06,
+      "loss": 5.8164,
+      "step": 545000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 9.121108175089039e-06,
+      "loss": 5.8126,
+      "step": 545500
+    },
+    {
+      "epoch": 8.94,
+      "step": 545561,
+      "total_flos": 3.6753587237709414e+17,
+      "train_loss": 6.136735617761035,
+      "train_runtime": 197999.1773,
+      "train_samples_per_second": 2958.981,
+      "train_steps_per_second": 30.823
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 6102900,
+  "num_train_epochs": 100,
+  "save_steps": 1000000,
+  "total_flos": 3.6753587237709414e+17,
+  "trial_name": null,
+  "trial_params": null
+}