diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8752 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9995057389203141,
+  "global_step": 728000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9965676313910704e-05,
+      "loss": 2.299,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9931352627821406e-05,
+      "loss": 2.1716,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989702894173211e-05,
+      "loss": 2.083,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986270525564282e-05,
+      "loss": 2.04,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982838156955352e-05,
+      "loss": 2.0074,
+      "step": 2500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.979405788346422e-05,
+      "loss": 1.9771,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975973419737492e-05,
+      "loss": 1.9466,
+      "step": 3500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972541051128563e-05,
+      "loss": 1.9384,
+      "step": 4000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9691086825196334e-05,
+      "loss": 1.9446,
+      "step": 4500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9656763139107036e-05,
+      "loss": 1.9054,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962243945301774e-05,
+      "loss": 1.8931,
+      "step": 5500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958811576692845e-05,
+      "loss": 1.8744,
+      "step": 6000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955379208083915e-05,
+      "loss": 1.8849,
+      "step": 6500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951946839474985e-05,
+      "loss": 1.8659,
+      "step": 7000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948514470866055e-05,
+      "loss": 1.8463,
+      "step": 7500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945082102257126e-05,
+      "loss": 1.8243,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9416497336481957e-05,
+      "loss": 1.8247,
+      "step": 8500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9382173650392665e-05,
+      "loss": 1.8389,
+      "step": 9000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934784996430337e-05,
+      "loss": 1.8103,
+      "step": 9500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9313526278214076e-05,
+      "loss": 1.7774,
+      "step": 10000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927920259212477e-05,
+      "loss": 1.7758,
+      "step": 10500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924487890603548e-05,
+      "loss": 1.7937,
+      "step": 11000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921055521994618e-05,
+      "loss": 1.7766,
+      "step": 11500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9176231533856884e-05,
+      "loss": 1.7564,
+      "step": 12000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9141907847767586e-05,
+      "loss": 1.7585,
+      "step": 12500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9107584161678295e-05,
+      "loss": 1.7528,
+      "step": 13000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9073260475589e-05,
+      "loss": 1.7505,
+      "step": 13500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90389367894997e-05,
+      "loss": 1.7548,
+      "step": 14000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90046131034104e-05,
+      "loss": 1.7404,
+      "step": 14500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897028941732111e-05,
+      "loss": 1.7278,
+      "step": 15000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893596573123181e-05,
+      "loss": 1.7325,
+      "step": 15500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8901642045142514e-05,
+      "loss": 1.7148,
+      "step": 16000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8867318359053216e-05,
+      "loss": 1.7168,
+      "step": 16500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8832994672963924e-05,
+      "loss": 1.7354,
+      "step": 17000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879867098687462e-05,
+      "loss": 1.712,
+      "step": 17500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876434730078533e-05,
+      "loss": 1.7054,
+      "step": 18000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.873002361469603e-05,
+      "loss": 1.7183,
+      "step": 18500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.869569992860674e-05,
+      "loss": 1.6779,
+      "step": 19000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8661376242517434e-05,
+      "loss": 1.7045,
+      "step": 19500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862705255642814e-05,
+      "loss": 1.6761,
+      "step": 20000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8592728870338845e-05,
+      "loss": 1.6818,
+      "step": 20500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.855840518424955e-05,
+      "loss": 1.6836,
+      "step": 21000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.852408149816025e-05,
+      "loss": 1.6878,
+      "step": 21500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.848975781207096e-05,
+      "loss": 1.6763,
+      "step": 22000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.845543412598166e-05,
+      "loss": 1.6657,
+      "step": 22500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.842111043989236e-05,
+      "loss": 1.6568,
+      "step": 23000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8386786753803064e-05,
+      "loss": 1.6665,
+      "step": 23500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.835246306771377e-05,
+      "loss": 1.6809,
+      "step": 24000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8318139381624475e-05,
+      "loss": 1.6569,
+      "step": 24500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8283815695535177e-05,
+      "loss": 1.6375,
+      "step": 25000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.824949200944588e-05,
+      "loss": 1.6595,
+      "step": 25500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.821516832335659e-05,
+      "loss": 1.6409,
+      "step": 26000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.818084463726729e-05,
+      "loss": 1.6374,
+      "step": 26500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.814652095117799e-05,
+      "loss": 1.6294,
+      "step": 27000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.811219726508869e-05,
+      "loss": 1.633,
+      "step": 27500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80778735789994e-05,
+      "loss": 1.6388,
+      "step": 28000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80435498929101e-05,
+      "loss": 1.6359,
+      "step": 28500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8009226206820806e-05,
+      "loss": 1.6297,
+      "step": 29000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.797490252073151e-05,
+      "loss": 1.635,
+      "step": 29500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.794057883464222e-05,
+      "loss": 1.629,
+      "step": 30000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.790625514855291e-05,
+      "loss": 1.609,
+      "step": 30500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.787193146246362e-05,
+      "loss": 1.6163,
+      "step": 31000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.783760777637432e-05,
+      "loss": 1.6199,
+      "step": 31500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7803284090285025e-05,
+      "loss": 1.5975,
+      "step": 32000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.776896040419573e-05,
+      "loss": 1.626,
+      "step": 32500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7734636718106436e-05,
+      "loss": 1.6036,
+      "step": 33000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.770031303201714e-05,
+      "loss": 1.6077,
+      "step": 33500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766598934592784e-05,
+      "loss": 1.6072,
+      "step": 34000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763166565983854e-05,
+      "loss": 1.5979,
+      "step": 34500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759734197374925e-05,
+      "loss": 1.6057,
+      "step": 35000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756301828765995e-05,
+      "loss": 1.5918,
+      "step": 35500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7528694601570654e-05,
+      "loss": 1.5925,
+      "step": 36000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7494370915481356e-05,
+      "loss": 1.6056,
+      "step": 36500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7460047229392065e-05,
+      "loss": 1.5938,
+      "step": 37000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.742572354330276e-05,
+      "loss": 1.5799,
+      "step": 37500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.739139985721347e-05,
+      "loss": 1.5874,
+      "step": 38000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.735707617112417e-05,
+      "loss": 1.5895,
+      "step": 38500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.732275248503488e-05,
+      "loss": 1.5943,
+      "step": 39000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7288428798945575e-05,
+      "loss": 1.5695,
+      "step": 39500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7254105112856284e-05,
+      "loss": 1.5784,
+      "step": 40000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7219781426766986e-05,
+      "loss": 1.5821,
+      "step": 40500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.718545774067769e-05,
+      "loss": 1.5812,
+      "step": 41000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.715113405458839e-05,
+      "loss": 1.5745,
+      "step": 41500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71168103684991e-05,
+      "loss": 1.5725,
+      "step": 42000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70824866824098e-05,
+      "loss": 1.5708,
+      "step": 42500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70481629963205e-05,
+      "loss": 1.5654,
+      "step": 43000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7013839310231204e-05,
+      "loss": 1.5728,
+      "step": 43500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.697951562414191e-05,
+      "loss": 1.5469,
+      "step": 44000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6945191938052615e-05,
+      "loss": 1.5794,
+      "step": 44500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.691086825196332e-05,
+      "loss": 1.5555,
+      "step": 45000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.687654456587402e-05,
+      "loss": 1.5621,
+      "step": 45500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.684222087978473e-05,
+      "loss": 1.575,
+      "step": 46000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.680789719369543e-05,
+      "loss": 1.547,
+      "step": 46500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.677357350760613e-05,
+      "loss": 1.538,
+      "step": 47000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6739249821516834e-05,
+      "loss": 1.5478,
+      "step": 47500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.670492613542754e-05,
+      "loss": 1.5501,
+      "step": 48000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.667060244933824e-05,
+      "loss": 1.545,
+      "step": 48500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.663627876324895e-05,
+      "loss": 1.5574,
+      "step": 49000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.660195507715965e-05,
+      "loss": 1.5398,
+      "step": 49500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.656763139107036e-05,
+      "loss": 1.5406,
+      "step": 50000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.653330770498105e-05,
+      "loss": 1.5354,
+      "step": 50500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.649898401889176e-05,
+      "loss": 1.5448,
+      "step": 51000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6464660332802463e-05,
+      "loss": 1.5261,
+      "step": 51500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6430336646713165e-05,
+      "loss": 1.5385,
+      "step": 52000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.639601296062387e-05,
+      "loss": 1.5363,
+      "step": 52500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6361689274534576e-05,
+      "loss": 1.5315,
+      "step": 53000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.632736558844528e-05,
+      "loss": 1.5351,
+      "step": 53500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.629304190235598e-05,
+      "loss": 1.5171,
+      "step": 54000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.625871821626668e-05,
+      "loss": 1.5275,
+      "step": 54500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.622439453017739e-05,
+      "loss": 1.5402,
+      "step": 55000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.619007084408809e-05,
+      "loss": 1.5287,
+      "step": 55500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6155747157998795e-05,
+      "loss": 1.5232,
+      "step": 56000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61214234719095e-05,
+      "loss": 1.538,
+      "step": 56500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6087099785820206e-05,
+      "loss": 1.522,
+      "step": 57000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.605277609973091e-05,
+      "loss": 1.5132,
+      "step": 57500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601845241364161e-05,
+      "loss": 1.5273,
+      "step": 58000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.598412872755231e-05,
+      "loss": 1.509,
+      "step": 58500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.594980504146302e-05,
+      "loss": 1.5257,
+      "step": 59000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5915481355373716e-05,
+      "loss": 1.5165,
+      "step": 59500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5881157669284424e-05,
+      "loss": 1.5089,
+      "step": 60000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5846833983195126e-05,
+      "loss": 1.5302,
+      "step": 60500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.581251029710583e-05,
+      "loss": 1.5085,
+      "step": 61000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.577818661101653e-05,
+      "loss": 1.5038,
+      "step": 61500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.574386292492723e-05,
+      "loss": 1.5175,
+      "step": 62000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.570953923883794e-05,
+      "loss": 1.5259,
+      "step": 62500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.567521555274864e-05,
+      "loss": 1.4952,
+      "step": 63000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5640891866659345e-05,
+      "loss": 1.4924,
+      "step": 63500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.560656818057005e-05,
+      "loss": 1.4995,
+      "step": 64000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5572244494480756e-05,
+      "loss": 1.4919,
+      "step": 64500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.553792080839145e-05,
+      "loss": 1.5147,
+      "step": 65000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.550359712230216e-05,
+      "loss": 1.4928,
+      "step": 65500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.546927343621286e-05,
+      "loss": 1.4959,
+      "step": 66000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.543494975012357e-05,
+      "loss": 1.4969,
+      "step": 66500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5400626064034266e-05,
+      "loss": 1.5043,
+      "step": 67000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5366302377944975e-05,
+      "loss": 1.4929,
+      "step": 67500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.533197869185568e-05,
+      "loss": 1.4929,
+      "step": 68000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.529765500576638e-05,
+      "loss": 1.4884,
+      "step": 68500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.526333131967708e-05,
+      "loss": 1.486,
+      "step": 69000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.522900763358779e-05,
+      "loss": 1.4958,
+      "step": 69500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.519468394749849e-05,
+      "loss": 1.4976,
+      "step": 70000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.516036026140919e-05,
+      "loss": 1.4888,
+      "step": 70500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5126036575319895e-05,
+      "loss": 1.4877,
+      "step": 71000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5091712889230604e-05,
+      "loss": 1.4691,
+      "step": 71500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5057389203141306e-05,
+      "loss": 1.4788,
+      "step": 72000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.502306551705201e-05,
+      "loss": 1.4976,
+      "step": 72500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.498874183096271e-05,
+      "loss": 1.499,
+      "step": 73000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.495441814487342e-05,
+      "loss": 1.4825,
+      "step": 73500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.492009445878412e-05,
+      "loss": 1.4849,
+      "step": 74000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.488577077269482e-05,
+      "loss": 1.4852,
+      "step": 74500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4851447086605525e-05,
+      "loss": 1.488,
+      "step": 75000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4817123400516234e-05,
+      "loss": 1.4779,
+      "step": 75500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.478279971442693e-05,
+      "loss": 1.4731,
+      "step": 76000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.474847602833764e-05,
+      "loss": 1.4766,
+      "step": 76500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.471415234224834e-05,
+      "loss": 1.498,
+      "step": 77000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.467982865615905e-05,
+      "loss": 1.4696,
+      "step": 77500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4645504970069744e-05,
+      "loss": 1.4849,
+      "step": 78000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.461118128398045e-05,
+      "loss": 1.4802,
+      "step": 78500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4576857597891154e-05,
+      "loss": 1.4857,
+      "step": 79000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4542533911801856e-05,
+      "loss": 1.4691,
+      "step": 79500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.450821022571256e-05,
+      "loss": 1.4791,
+      "step": 80000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.447388653962327e-05,
+      "loss": 1.4448,
+      "step": 80500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.443956285353397e-05,
+      "loss": 1.4589,
+      "step": 81000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.440523916744467e-05,
+      "loss": 1.4681,
+      "step": 81500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.437091548135537e-05,
+      "loss": 1.474,
+      "step": 82000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.433659179526608e-05,
+      "loss": 1.4692,
+      "step": 82500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4302268109176784e-05,
+      "loss": 1.4631,
+      "step": 83000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4267944423087486e-05,
+      "loss": 1.4745,
+      "step": 83500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.423362073699819e-05,
+      "loss": 1.4657,
+      "step": 84000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4199297050908897e-05,
+      "loss": 1.4642,
+      "step": 84500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.416497336481959e-05,
+      "loss": 1.4688,
+      "step": 85000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.41306496787303e-05,
+      "loss": 1.4422,
+      "step": 85500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4096325992641e-05,
+      "loss": 1.4623,
+      "step": 86000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.406200230655171e-05,
+      "loss": 1.4744,
+      "step": 86500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4027678620462407e-05,
+      "loss": 1.4449,
+      "step": 87000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3993354934373115e-05,
+      "loss": 1.4662,
+      "step": 87500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.395903124828382e-05,
+      "loss": 1.4489,
+      "step": 88000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.392470756219452e-05,
+      "loss": 1.4465,
+      "step": 88500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.389038387610522e-05,
+      "loss": 1.4508,
+      "step": 89000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.385606019001593e-05,
+      "loss": 1.4566,
+      "step": 89500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.382173650392663e-05,
+      "loss": 1.4495,
+      "step": 90000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3787412817837334e-05,
+      "loss": 1.4547,
+      "step": 90500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3753089131748036e-05,
+      "loss": 1.4525,
+      "step": 91000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3718765445658745e-05,
+      "loss": 1.4435,
+      "step": 91500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368444175956945e-05,
+      "loss": 1.4497,
+      "step": 92000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.365011807348015e-05,
+      "loss": 1.4472,
+      "step": 92500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.361579438739085e-05,
+      "loss": 1.4556,
+      "step": 93000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358147070130156e-05,
+      "loss": 1.4483,
+      "step": 93500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.354714701521226e-05,
+      "loss": 1.4345,
+      "step": 94000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3512823329122964e-05,
+      "loss": 1.4402,
+      "step": 94500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3478499643033666e-05,
+      "loss": 1.4376,
+      "step": 95000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3444175956944374e-05,
+      "loss": 1.4474,
+      "step": 95500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.340985227085507e-05,
+      "loss": 1.439,
+      "step": 96000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.337552858476578e-05,
+      "loss": 1.4424,
+      "step": 96500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.334120489867648e-05,
+      "loss": 1.4493,
+      "step": 97000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.330688121258719e-05,
+      "loss": 1.4319,
+      "step": 97500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3272557526497884e-05,
+      "loss": 1.4312,
+      "step": 98000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.323823384040859e-05,
+      "loss": 1.4547,
+      "step": 98500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3203910154319295e-05,
+      "loss": 1.4464,
+      "step": 99000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.316958646823e-05,
+      "loss": 1.4292,
+      "step": 99500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.31352627821407e-05,
+      "loss": 1.4199,
+      "step": 100000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.310093909605141e-05,
+      "loss": 1.4278,
+      "step": 100500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.306661540996211e-05,
+      "loss": 1.4218,
+      "step": 101000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303229172387281e-05,
+      "loss": 1.4267,
+      "step": 101500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2997968037783514e-05,
+      "loss": 1.4229,
+      "step": 102000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.296364435169422e-05,
+      "loss": 1.4245,
+      "step": 102500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2929320665604925e-05,
+      "loss": 1.4369,
+      "step": 103000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2894996979515627e-05,
+      "loss": 1.4236,
+      "step": 103500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.286067329342633e-05,
+      "loss": 1.4441,
+      "step": 104000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.282634960733704e-05,
+      "loss": 1.43,
+      "step": 104500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.279202592124774e-05,
+      "loss": 1.427,
+      "step": 105000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.275770223515844e-05,
+      "loss": 1.4124,
+      "step": 105500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.272337854906914e-05,
+      "loss": 1.4155,
+      "step": 106000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.268905486297985e-05,
+      "loss": 1.4292,
+      "step": 106500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.265473117689055e-05,
+      "loss": 1.4286,
+      "step": 107000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2620407490801256e-05,
+      "loss": 1.4254,
+      "step": 107500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.258608380471196e-05,
+      "loss": 1.4151,
+      "step": 108000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.255176011862266e-05,
+      "loss": 1.4189,
+      "step": 108500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.251743643253336e-05,
+      "loss": 1.427,
+      "step": 109000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.248311274644407e-05,
+      "loss": 1.4256,
+      "step": 109500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.244878906035477e-05,
+      "loss": 1.4059,
+      "step": 110000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2414465374265475e-05,
+      "loss": 1.4165,
+      "step": 110500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.238014168817618e-05,
+      "loss": 1.4177,
+      "step": 111000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2345818002086885e-05,
+      "loss": 1.4236,
+      "step": 111500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.231149431599759e-05,
+      "loss": 1.4325,
+      "step": 112000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.227717062990829e-05,
+      "loss": 1.4078,
+      "step": 112500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.224284694381899e-05,
+      "loss": 1.4134,
+      "step": 113000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.22085232577297e-05,
+      "loss": 1.4013,
+      "step": 113500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.21741995716404e-05,
+      "loss": 1.4128,
+      "step": 114000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2139875885551104e-05,
+      "loss": 1.4254,
+      "step": 114500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2105552199461806e-05,
+      "loss": 1.4164,
+      "step": 115000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2071228513372515e-05,
+      "loss": 1.422,
+      "step": 115500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.203690482728321e-05,
+      "loss": 1.4296,
+      "step": 116000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.200258114119392e-05,
+      "loss": 1.4151,
+      "step": 116500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.196825745510462e-05,
+      "loss": 1.3993,
+      "step": 117000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.193393376901533e-05,
+      "loss": 1.4046,
+      "step": 117500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1899610082926025e-05,
+      "loss": 1.4134,
+      "step": 118000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1865286396836734e-05,
+      "loss": 1.3971,
+      "step": 118500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1830962710747436e-05,
+      "loss": 1.4095,
+      "step": 119000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.179663902465814e-05,
+      "loss": 1.3892,
+      "step": 119500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.176231533856884e-05,
+      "loss": 1.3881,
+      "step": 120000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.172799165247955e-05,
+      "loss": 1.4076,
+      "step": 120500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.169366796639025e-05,
+      "loss": 1.3896,
+      "step": 121000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.165934428030095e-05,
+      "loss": 1.4049,
+      "step": 121500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1625020594211654e-05,
+      "loss": 1.3841,
+      "step": 122000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1590696908122356e-05,
+      "loss": 1.4008,
+      "step": 122500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1556373222033065e-05,
+      "loss": 1.3976,
+      "step": 123000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.152204953594376e-05,
+      "loss": 1.3941,
+      "step": 123500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.148772584985447e-05,
+      "loss": 1.3983,
+      "step": 124000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.145340216376517e-05,
+      "loss": 1.3901,
+      "step": 124500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.141907847767588e-05,
+      "loss": 1.4039,
+      "step": 125000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1384754791586575e-05,
+      "loss": 1.3999,
+      "step": 125500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1350431105497284e-05,
+      "loss": 1.3893,
+      "step": 126000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1316107419407986e-05,
+      "loss": 1.3875,
+      "step": 126500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.128178373331869e-05,
+      "loss": 1.4075,
+      "step": 127000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.124746004722939e-05,
+      "loss": 1.4106,
+      "step": 127500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.12131363611401e-05,
+      "loss": 1.3838,
+      "step": 128000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11788126750508e-05,
+      "loss": 1.3982,
+      "step": 128500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11444889889615e-05,
+      "loss": 1.4009,
+      "step": 129000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1110165302872205e-05,
+      "loss": 1.3753,
+      "step": 129500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1075841616782913e-05,
+      "loss": 1.379,
+      "step": 130000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1041517930693615e-05,
+      "loss": 1.3839,
+      "step": 130500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.100719424460432e-05,
+      "loss": 1.3887,
+      "step": 131000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.097287055851502e-05,
+      "loss": 1.377,
+      "step": 131500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.093854687242573e-05,
+      "loss": 1.3908,
+      "step": 132000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.090422318633642e-05,
+      "loss": 1.394,
+      "step": 132500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.086989950024713e-05,
+      "loss": 1.4083,
+      "step": 133000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0835575814157834e-05,
+      "loss": 1.3718,
+      "step": 133500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.080125212806854e-05,
+      "loss": 1.3849,
+      "step": 134000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.076692844197924e-05,
+      "loss": 1.3941,
+      "step": 134500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.073260475588995e-05,
+      "loss": 1.3769,
+      "step": 135000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069828106980065e-05,
+      "loss": 1.3808,
+      "step": 135500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.066395738371135e-05,
+      "loss": 1.3851,
+      "step": 136000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.062963369762205e-05,
+      "loss": 1.3785,
+      "step": 136500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.059531001153276e-05,
+      "loss": 1.3727,
+      "step": 137000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0560986325443464e-05,
+      "loss": 1.3997,
+      "step": 137500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0526662639354166e-05,
+      "loss": 1.3763,
+      "step": 138000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.049233895326487e-05,
+      "loss": 1.3935,
+      "step": 138500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0458015267175576e-05,
+      "loss": 1.3733,
+      "step": 139000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.042369158108628e-05,
+      "loss": 1.3802,
+      "step": 139500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.038936789499698e-05,
+      "loss": 1.369,
+      "step": 140000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.035504420890768e-05,
+      "loss": 1.3751,
+      "step": 140500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.032072052281839e-05,
+      "loss": 1.3664,
+      "step": 141000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.028639683672909e-05,
+      "loss": 1.3892,
+      "step": 141500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0252073150639795e-05,
+      "loss": 1.3661,
+      "step": 142000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.02177494645505e-05,
+      "loss": 1.3708,
+      "step": 142500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0183425778461206e-05,
+      "loss": 1.3713,
+      "step": 143000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01491020923719e-05,
+      "loss": 1.3754,
+      "step": 143500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.011477840628261e-05,
+      "loss": 1.3628,
+      "step": 144000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.008045472019331e-05,
+      "loss": 1.372,
+      "step": 144500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.004613103410402e-05,
+      "loss": 1.3484,
+      "step": 145000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0011807348014716e-05,
+      "loss": 1.3768,
+      "step": 145500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9977483661925425e-05,
+      "loss": 1.3821,
+      "step": 146000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9943159975836127e-05,
+      "loss": 1.3691,
+      "step": 146500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.990883628974683e-05,
+      "loss": 1.3641,
+      "step": 147000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.987451260365753e-05,
+      "loss": 1.368,
+      "step": 147500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.984018891756824e-05,
+      "loss": 1.3663,
+      "step": 148000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.980586523147894e-05,
+      "loss": 1.3698,
+      "step": 148500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.977154154538964e-05,
+      "loss": 1.3784,
+      "step": 149000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9737217859300345e-05,
+      "loss": 1.3594,
+      "step": 149500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9702894173211054e-05,
+      "loss": 1.3599,
+      "step": 150000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9668570487121756e-05,
+      "loss": 1.38,
+      "step": 150500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.963424680103246e-05,
+      "loss": 1.3421,
+      "step": 151000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959992311494316e-05,
+      "loss": 1.3661,
+      "step": 151500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956559942885387e-05,
+      "loss": 1.3608,
+      "step": 152000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.953127574276457e-05,
+      "loss": 1.368,
+      "step": 152500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.949695205667527e-05,
+      "loss": 1.3556,
+      "step": 153000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9462628370585975e-05,
+      "loss": 1.3744,
+      "step": 153500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9428304684496684e-05,
+      "loss": 1.359,
+      "step": 154000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.939398099840738e-05,
+      "loss": 1.3587,
+      "step": 154500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.935965731231809e-05,
+      "loss": 1.3647,
+      "step": 155000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.932533362622879e-05,
+      "loss": 1.3677,
+      "step": 155500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.929100994013949e-05,
+      "loss": 1.3602,
+      "step": 156000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9256686254050194e-05,
+      "loss": 1.3743,
+      "step": 156500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.92223625679609e-05,
+      "loss": 1.3637,
+      "step": 157000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9188038881871604e-05,
+      "loss": 1.3753,
+      "step": 157500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9153715195782306e-05,
+      "loss": 1.3624,
+      "step": 158000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.911939150969301e-05,
+      "loss": 1.3525,
+      "step": 158500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.908506782360372e-05,
+      "loss": 1.3484,
+      "step": 159000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.905074413751442e-05,
+      "loss": 1.3584,
+      "step": 159500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.901642045142512e-05,
+      "loss": 1.3617,
+      "step": 160000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.898209676533582e-05,
+      "loss": 1.3553,
+      "step": 160500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894777307924653e-05,
+      "loss": 1.3524,
+      "step": 161000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8913449393157234e-05,
+      "loss": 1.3708,
+      "step": 161500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8879125707067936e-05,
+      "loss": 1.3372,
+      "step": 162000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.884480202097864e-05,
+      "loss": 1.3572,
+      "step": 162500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8810478334889347e-05,
+      "loss": 1.3373,
+      "step": 163000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.877615464880004e-05,
+      "loss": 1.3364,
+      "step": 163500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.874183096271075e-05,
+      "loss": 1.3539,
+      "step": 164000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.870750727662145e-05,
+      "loss": 1.3447,
+      "step": 164500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.867318359053216e-05,
+      "loss": 1.3547,
+      "step": 165000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8638859904442857e-05,
+      "loss": 1.3542,
+      "step": 165500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8604536218353565e-05,
+      "loss": 1.3433,
+      "step": 166000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.857021253226427e-05,
+      "loss": 1.3443,
+      "step": 166500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.853588884617497e-05,
+      "loss": 1.3488,
+      "step": 167000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.850156516008567e-05,
+      "loss": 1.3495,
+      "step": 167500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.846724147399638e-05,
+      "loss": 1.3535,
+      "step": 168000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.843291778790708e-05,
+      "loss": 1.3285,
+      "step": 168500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8398594101817784e-05,
+      "loss": 1.3522,
+      "step": 169000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8364270415728486e-05,
+      "loss": 1.3613,
+      "step": 169500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8329946729639195e-05,
+      "loss": 1.3612,
+      "step": 170000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.82956230435499e-05,
+      "loss": 1.339,
+      "step": 170500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.82612993574606e-05,
+      "loss": 1.3508,
+      "step": 171000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.82269756713713e-05,
+      "loss": 1.3573,
+      "step": 171500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.819265198528201e-05,
+      "loss": 1.3444,
+      "step": 172000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.815832829919271e-05,
+      "loss": 1.3544,
+      "step": 172500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8124004613103413e-05,
+      "loss": 1.3395,
+      "step": 173000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8089680927014115e-05,
+      "loss": 1.363,
+      "step": 173500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8055357240924824e-05,
+      "loss": 1.3448,
+      "step": 174000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.802103355483552e-05,
+      "loss": 1.339,
+      "step": 174500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.798670986874623e-05,
+      "loss": 1.3348,
+      "step": 175000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.795238618265693e-05,
+      "loss": 1.3315,
+      "step": 175500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.791806249656764e-05,
+      "loss": 1.3393,
+      "step": 176000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7883738810478334e-05,
+      "loss": 1.3353,
+      "step": 176500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.784941512438904e-05,
+      "loss": 1.3398,
+      "step": 177000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7815091438299745e-05,
+      "loss": 1.3396,
+      "step": 177500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.778076775221045e-05,
+      "loss": 1.3453,
+      "step": 178000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.774644406612115e-05,
+      "loss": 1.3311,
+      "step": 178500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.771212038003186e-05,
+      "loss": 1.3492,
+      "step": 179000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.767779669394256e-05,
+      "loss": 1.32,
+      "step": 179500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.764347300785326e-05,
+      "loss": 1.3049,
+      "step": 180000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7609149321763964e-05,
+      "loss": 1.3506,
+      "step": 180500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.757482563567467e-05,
+      "loss": 1.3235,
+      "step": 181000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7540501949585374e-05,
+      "loss": 1.3252,
+      "step": 181500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7506178263496076e-05,
+      "loss": 1.3228,
+      "step": 182000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.747185457740678e-05,
+      "loss": 1.3312,
+      "step": 182500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.743753089131749e-05,
+      "loss": 1.343,
+      "step": 183000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.740320720522818e-05,
+      "loss": 1.3254,
+      "step": 183500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7368883519138884e-05,
+      "loss": 1.3474,
+      "step": 184000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.733455983304959e-05,
+      "loss": 1.3312,
+      "step": 184500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7300236146960295e-05,
+      "loss": 1.3366,
+      "step": 185000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7265912460871e-05,
+      "loss": 1.339,
+      "step": 185500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.72315887747817e-05,
+      "loss": 1.3227,
+      "step": 186000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.719726508869241e-05,
+      "loss": 1.3249,
+      "step": 186500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.716294140260311e-05,
+      "loss": 1.3215,
+      "step": 187000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.712861771651381e-05,
+      "loss": 1.3249,
+      "step": 187500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7094294030424514e-05,
+      "loss": 1.3297,
+      "step": 188000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.705997034433522e-05,
+      "loss": 1.3154,
+      "step": 188500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7025646658245925e-05,
+      "loss": 1.3291,
+      "step": 189000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.699132297215663e-05,
+      "loss": 1.3347,
+      "step": 189500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.695699928606733e-05,
+      "loss": 1.3247,
+      "step": 190000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.692267559997804e-05,
+      "loss": 1.3304,
+      "step": 190500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.688835191388873e-05,
+      "loss": 1.3286,
+      "step": 191000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.685402822779944e-05,
+      "loss": 1.3131,
+      "step": 191500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6819704541710143e-05,
+      "loss": 1.3191,
+      "step": 192000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.678538085562085e-05,
+      "loss": 1.3129,
+      "step": 192500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.675105716953155e-05,
+      "loss": 1.3022,
+      "step": 193000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6716733483442256e-05,
+      "loss": 1.332,
+      "step": 193500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.668240979735296e-05,
+      "loss": 1.3169,
+      "step": 194000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.664808611126366e-05,
+      "loss": 1.3113,
+      "step": 194500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.661376242517436e-05,
+      "loss": 1.325,
+      "step": 195000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.657943873908507e-05,
+      "loss": 1.3363,
+      "step": 195500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.654511505299577e-05,
+      "loss": 1.3304,
+      "step": 196000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6510791366906475e-05,
+      "loss": 1.3147,
+      "step": 196500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.647646768081718e-05,
+      "loss": 1.3202,
+      "step": 197000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6442143994727886e-05,
+      "loss": 1.3182,
+      "step": 197500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.640782030863859e-05,
+      "loss": 1.3251,
+      "step": 198000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.637349662254929e-05,
+      "loss": 1.3317,
+      "step": 198500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.633917293645999e-05,
+      "loss": 1.3264,
+      "step": 199000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63048492503707e-05,
+      "loss": 1.3281,
+      "step": 199500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.62705255642814e-05,
+      "loss": 1.309,
+      "step": 200000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6236201878192104e-05,
+      "loss": 1.323,
+      "step": 200500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6201878192102806e-05,
+      "loss": 1.315,
+      "step": 201000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6167554506013515e-05,
+      "loss": 1.3242,
+      "step": 201500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.613323081992421e-05,
+      "loss": 1.3092,
+      "step": 202000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.609890713383492e-05,
+      "loss": 1.3044,
+      "step": 202500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.606458344774562e-05,
+      "loss": 1.3132,
+      "step": 203000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.603025976165632e-05,
+      "loss": 1.3191,
+      "step": 203500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5995936075567025e-05,
+      "loss": 1.3096,
+      "step": 204000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5961612389477734e-05,
+      "loss": 1.2949,
+      "step": 204500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5927288703388436e-05,
+      "loss": 1.3083,
+      "step": 205000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.589296501729914e-05,
+      "loss": 1.2998,
+      "step": 205500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.585864133120984e-05,
+      "loss": 1.3158,
+      "step": 206000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.582431764512055e-05,
+      "loss": 1.2995,
+      "step": 206500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.578999395903125e-05,
+      "loss": 1.3204,
+      "step": 207000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.575567027294195e-05,
+      "loss": 1.3131,
+      "step": 207500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5721346586852655e-05,
+      "loss": 1.2995,
+      "step": 208000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.568702290076336e-05,
+      "loss": 1.3058,
+      "step": 208500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5652699214674065e-05,
+      "loss": 1.3073,
+      "step": 209000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.561837552858477e-05,
+      "loss": 1.3096,
+      "step": 209500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.558405184249547e-05,
+      "loss": 1.3014,
+      "step": 210000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.554972815640618e-05,
+      "loss": 1.3017,
+      "step": 210500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.551540447031687e-05,
+      "loss": 1.2992,
+      "step": 211000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.548108078422758e-05,
+      "loss": 1.3256,
+      "step": 211500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5446757098138284e-05,
+      "loss": 1.3012,
+      "step": 212000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.541243341204899e-05,
+      "loss": 1.2812,
+      "step": 212500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.537810972595969e-05,
+      "loss": 1.3136,
+      "step": 213000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53437860398704e-05,
+      "loss": 1.3052,
+      "step": 213500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53094623537811e-05,
+      "loss": 1.2998,
+      "step": 214000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.52751386676918e-05,
+      "loss": 1.2986,
+      "step": 214500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.52408149816025e-05,
+      "loss": 1.3083,
+      "step": 215000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.520649129551321e-05,
+      "loss": 1.3027,
+      "step": 215500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5172167609423914e-05,
+      "loss": 1.2866,
+      "step": 216000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5137843923334616e-05,
+      "loss": 1.2938,
+      "step": 216500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.510352023724532e-05,
+      "loss": 1.3157,
+      "step": 217000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5069196551156026e-05,
+      "loss": 1.286,
+      "step": 217500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.503487286506673e-05,
+      "loss": 1.3145,
+      "step": 218000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.500054917897743e-05,
+      "loss": 1.2871,
+      "step": 218500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.496622549288813e-05,
+      "loss": 1.2767,
+      "step": 219000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493190180679884e-05,
+      "loss": 1.3071,
+      "step": 219500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.489757812070954e-05,
+      "loss": 1.2969,
+      "step": 220000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4863254434620245e-05,
+      "loss": 1.3044,
+      "step": 220500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.482893074853095e-05,
+      "loss": 1.3107,
+      "step": 221000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4794607062441656e-05,
+      "loss": 1.294,
+      "step": 221500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.476028337635235e-05,
+      "loss": 1.2912,
+      "step": 222000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.472595969026306e-05,
+      "loss": 1.3051,
+      "step": 222500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.469163600417376e-05,
+      "loss": 1.3039,
+      "step": 223000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.465731231808447e-05,
+      "loss": 1.2867,
+      "step": 223500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4622988631995166e-05,
+      "loss": 1.286,
+      "step": 224000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4588664945905875e-05,
+      "loss": 1.3003,
+      "step": 224500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4554341259816577e-05,
+      "loss": 1.3107,
+      "step": 225000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.452001757372728e-05,
+      "loss": 1.2988,
+      "step": 225500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.448569388763798e-05,
+      "loss": 1.2908,
+      "step": 226000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.445137020154869e-05,
+      "loss": 1.302,
+      "step": 226500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.441704651545939e-05,
+      "loss": 1.2844,
+      "step": 227000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.438272282937009e-05,
+      "loss": 1.2787,
+      "step": 227500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4348399143280795e-05,
+      "loss": 1.3059,
+      "step": 228000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4314075457191504e-05,
+      "loss": 1.3013,
+      "step": 228500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4279751771102206e-05,
+      "loss": 1.3059,
+      "step": 229000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.424542808501291e-05,
+      "loss": 1.2924,
+      "step": 229500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.421110439892361e-05,
+      "loss": 1.2929,
+      "step": 230000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.417678071283432e-05,
+      "loss": 1.2897,
+      "step": 230500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4142457026745014e-05,
+      "loss": 1.3013,
+      "step": 231000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410813334065572e-05,
+      "loss": 1.2833,
+      "step": 231500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4073809654566425e-05,
+      "loss": 1.272,
+      "step": 232000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4039485968477134e-05,
+      "loss": 1.2829,
+      "step": 232500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.400516228238783e-05,
+      "loss": 1.2853,
+      "step": 233000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.397083859629854e-05,
+      "loss": 1.2925,
+      "step": 233500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.393651491020924e-05,
+      "loss": 1.3131,
+      "step": 234000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.390219122411994e-05,
+      "loss": 1.2929,
+      "step": 234500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3867867538030643e-05,
+      "loss": 1.2761,
+      "step": 235000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.383354385194135e-05,
+      "loss": 1.2832,
+      "step": 235500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3799220165852054e-05,
+      "loss": 1.2793,
+      "step": 236000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3764896479762756e-05,
+      "loss": 1.282,
+      "step": 236500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.373057279367346e-05,
+      "loss": 1.2848,
+      "step": 237000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.369624910758417e-05,
+      "loss": 1.3026,
+      "step": 237500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.366192542149487e-05,
+      "loss": 1.2793,
+      "step": 238000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.362760173540557e-05,
+      "loss": 1.2878,
+      "step": 238500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.359327804931627e-05,
+      "loss": 1.2957,
+      "step": 239000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.355895436322698e-05,
+      "loss": 1.2903,
+      "step": 239500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3524630677137684e-05,
+      "loss": 1.2715,
+      "step": 240000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3490306991048386e-05,
+      "loss": 1.2904,
+      "step": 240500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.345598330495909e-05,
+      "loss": 1.2847,
+      "step": 241000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3421659618869796e-05,
+      "loss": 1.2684,
+      "step": 241500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.338733593278049e-05,
+      "loss": 1.2786,
+      "step": 242000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33530122466912e-05,
+      "loss": 1.2937,
+      "step": 242500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33186885606019e-05,
+      "loss": 1.2848,
+      "step": 243000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.328436487451261e-05,
+      "loss": 1.2762,
+      "step": 243500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3250041188423306e-05,
+      "loss": 1.288,
+      "step": 244000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.321571750233401e-05,
+      "loss": 1.2726,
+      "step": 244500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.318139381624472e-05,
+      "loss": 1.2991,
+      "step": 245000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.314707013015542e-05,
+      "loss": 1.2633,
+      "step": 245500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.311274644406612e-05,
+      "loss": 1.2817,
+      "step": 246000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.307842275797682e-05,
+      "loss": 1.2862,
+      "step": 246500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.304409907188753e-05,
+      "loss": 1.267,
+      "step": 247000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3009775385798234e-05,
+      "loss": 1.2825,
+      "step": 247500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2975451699708936e-05,
+      "loss": 1.2701,
+      "step": 248000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294112801361964e-05,
+      "loss": 1.2977,
+      "step": 248500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.290680432753035e-05,
+      "loss": 1.2721,
+      "step": 249000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.287248064144104e-05,
+      "loss": 1.274,
+      "step": 249500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.283815695535175e-05,
+      "loss": 1.2937,
+      "step": 250000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.280383326926245e-05,
+      "loss": 1.2758,
+      "step": 250500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2769509583173155e-05,
+      "loss": 1.2683,
+      "step": 251000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.273518589708386e-05,
+      "loss": 1.2862,
+      "step": 251500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2700862210994565e-05,
+      "loss": 1.2713,
+      "step": 252000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.266653852490527e-05,
+      "loss": 1.2652,
+      "step": 252500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.263221483881597e-05,
+      "loss": 1.2665,
+      "step": 253000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.259789115272667e-05,
+      "loss": 1.2582,
+      "step": 253500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.256356746663738e-05,
+      "loss": 1.2558,
+      "step": 254000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.252924378054808e-05,
+      "loss": 1.2535,
+      "step": 254500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2494920094458784e-05,
+      "loss": 1.29,
+      "step": 255000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2460596408369486e-05,
+      "loss": 1.2836,
+      "step": 255500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2426272722280195e-05,
+      "loss": 1.2664,
+      "step": 256000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23919490361909e-05,
+      "loss": 1.2936,
+      "step": 256500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23576253501016e-05,
+      "loss": 1.2826,
+      "step": 257000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23233016640123e-05,
+      "loss": 1.2667,
+      "step": 257500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.228897797792301e-05,
+      "loss": 1.2803,
+      "step": 258000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2254654291833705e-05,
+      "loss": 1.2748,
+      "step": 258500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2220330605744414e-05,
+      "loss": 1.269,
+      "step": 259000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2186006919655116e-05,
+      "loss": 1.2711,
+      "step": 259500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2151683233565824e-05,
+      "loss": 1.2766,
+      "step": 260000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.211735954747652e-05,
+      "loss": 1.2703,
+      "step": 260500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.208303586138723e-05,
+      "loss": 1.267,
+      "step": 261000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.204871217529793e-05,
+      "loss": 1.2844,
+      "step": 261500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.201438848920863e-05,
+      "loss": 1.266,
+      "step": 262000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1980064803119334e-05,
+      "loss": 1.2532,
+      "step": 262500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.194574111703004e-05,
+      "loss": 1.2806,
+      "step": 263000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1911417430940745e-05,
+      "loss": 1.2638,
+      "step": 263500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.187709374485145e-05,
+      "loss": 1.2662,
+      "step": 264000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.184277005876215e-05,
+      "loss": 1.2638,
+      "step": 264500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.180844637267286e-05,
+      "loss": 1.2673,
+      "step": 265000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.177412268658356e-05,
+      "loss": 1.2537,
+      "step": 265500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.173979900049426e-05,
+      "loss": 1.258,
+      "step": 266000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1705475314404964e-05,
+      "loss": 1.2689,
+      "step": 266500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.167115162831567e-05,
+      "loss": 1.2554,
+      "step": 267000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1636827942226375e-05,
+      "loss": 1.2667,
+      "step": 267500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.160250425613708e-05,
+      "loss": 1.276,
+      "step": 268000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.156818057004778e-05,
+      "loss": 1.2716,
+      "step": 268500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.153385688395849e-05,
+      "loss": 1.265,
+      "step": 269000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.149953319786918e-05,
+      "loss": 1.2742,
+      "step": 269500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.146520951177989e-05,
+      "loss": 1.2617,
+      "step": 270000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143088582569059e-05,
+      "loss": 1.2666,
+      "step": 270500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13965621396013e-05,
+      "loss": 1.2465,
+      "step": 271000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1362238453512e-05,
+      "loss": 1.2708,
+      "step": 271500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1327914767422706e-05,
+      "loss": 1.2571,
+      "step": 272000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.129359108133341e-05,
+      "loss": 1.2547,
+      "step": 272500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.125926739524411e-05,
+      "loss": 1.2629,
+      "step": 273000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.122494370915481e-05,
+      "loss": 1.2648,
+      "step": 273500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.119062002306552e-05,
+      "loss": 1.2543,
+      "step": 274000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115629633697622e-05,
+      "loss": 1.2653,
+      "step": 274500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1121972650886925e-05,
+      "loss": 1.2709,
+      "step": 275000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.108764896479763e-05,
+      "loss": 1.2755,
+      "step": 275500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1053325278708336e-05,
+      "loss": 1.2653,
+      "step": 276000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.101900159261904e-05,
+      "loss": 1.249,
+      "step": 276500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.098467790652974e-05,
+      "loss": 1.2593,
+      "step": 277000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.095035422044044e-05,
+      "loss": 1.2533,
+      "step": 277500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.091603053435115e-05,
+      "loss": 1.2724,
+      "step": 278000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0881706848261846e-05,
+      "loss": 1.2583,
+      "step": 278500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0847383162172554e-05,
+      "loss": 1.2428,
+      "step": 279000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0813059476083256e-05,
+      "loss": 1.2461,
+      "step": 279500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0778735789993965e-05,
+      "loss": 1.2668,
+      "step": 280000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.074441210390466e-05,
+      "loss": 1.2578,
+      "step": 280500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.071008841781537e-05,
+      "loss": 1.2573,
+      "step": 281000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.067576473172607e-05,
+      "loss": 1.2494,
+      "step": 281500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.064144104563677e-05,
+      "loss": 1.2602,
+      "step": 282000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0607117359547475e-05,
+      "loss": 1.2584,
+      "step": 282500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0572793673458184e-05,
+      "loss": 1.261,
+      "step": 283000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0538469987368886e-05,
+      "loss": 1.2466,
+      "step": 283500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.050414630127959e-05,
+      "loss": 1.2482,
+      "step": 284000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.046982261519029e-05,
+      "loss": 1.2479,
+      "step": 284500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0435498929101e-05,
+      "loss": 1.2509,
+      "step": 285000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0401175243011697e-05,
+      "loss": 1.2528,
+      "step": 285500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0366851556922406e-05,
+      "loss": 1.2507,
+      "step": 286000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0332527870833105e-05,
+      "loss": 1.25,
+      "step": 286500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.029820418474381e-05,
+      "loss": 1.233,
+      "step": 287000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0263880498654512e-05,
+      "loss": 1.2248,
+      "step": 287500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0229556812565217e-05,
+      "loss": 1.2268,
+      "step": 288000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.019523312647592e-05,
+      "loss": 1.253,
+      "step": 288500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0160909440386625e-05,
+      "loss": 1.235,
+      "step": 289000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0126585754297327e-05,
+      "loss": 1.2512,
+      "step": 289500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0092262068208032e-05,
+      "loss": 1.2603,
+      "step": 290000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0057938382118734e-05,
+      "loss": 1.2679,
+      "step": 290500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.002361469602944e-05,
+      "loss": 1.2452,
+      "step": 291000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.998929100994014e-05,
+      "loss": 1.2507,
+      "step": 291500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9954967323850847e-05,
+      "loss": 1.2647,
+      "step": 292000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.992064363776155e-05,
+      "loss": 1.2478,
+      "step": 292500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9886319951672254e-05,
+      "loss": 1.2405,
+      "step": 293000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9851996265582953e-05,
+      "loss": 1.2553,
+      "step": 293500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.981767257949366e-05,
+      "loss": 1.2543,
+      "step": 294000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.978334889340436e-05,
+      "loss": 1.2524,
+      "step": 294500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.974902520731507e-05,
+      "loss": 1.2478,
+      "step": 295000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9714701521225768e-05,
+      "loss": 1.2429,
+      "step": 295500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9680377835136476e-05,
+      "loss": 1.2469,
+      "step": 296000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9646054149047175e-05,
+      "loss": 1.2442,
+      "step": 296500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.961173046295788e-05,
+      "loss": 1.2496,
+      "step": 297000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9577406776868582e-05,
+      "loss": 1.2361,
+      "step": 297500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9543083090779288e-05,
+      "loss": 1.2271,
+      "step": 298000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.950875940468999e-05,
+      "loss": 1.2513,
+      "step": 298500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9474435718600695e-05,
+      "loss": 1.2354,
+      "step": 299000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9440112032511397e-05,
+      "loss": 1.2414,
+      "step": 299500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9405788346422102e-05,
+      "loss": 1.2556,
+      "step": 300000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9371464660332804e-05,
+      "loss": 1.2525,
+      "step": 300500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.933714097424351e-05,
+      "loss": 1.2477,
+      "step": 301000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9302817288154212e-05,
+      "loss": 1.2258,
+      "step": 301500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9268493602064917e-05,
+      "loss": 1.2545,
+      "step": 302000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.923416991597562e-05,
+      "loss": 1.2424,
+      "step": 302500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9199846229886324e-05,
+      "loss": 1.2384,
+      "step": 303000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9165522543797023e-05,
+      "loss": 1.2437,
+      "step": 303500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9131198857707732e-05,
+      "loss": 1.2501,
+      "step": 304000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.909687517161843e-05,
+      "loss": 1.2394,
+      "step": 304500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.906255148552914e-05,
+      "loss": 1.2626,
+      "step": 305000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9028227799439838e-05,
+      "loss": 1.2452,
+      "step": 305500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.899390411335054e-05,
+      "loss": 1.255,
+      "step": 306000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8959580427261245e-05,
+      "loss": 1.2437,
+      "step": 306500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8925256741171947e-05,
+      "loss": 1.2505,
+      "step": 307000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8890933055082653e-05,
+      "loss": 1.2403,
+      "step": 307500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8856609368993355e-05,
+      "loss": 1.254,
+      "step": 308000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.882228568290406e-05,
+      "loss": 1.2384,
+      "step": 308500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8787961996814762e-05,
+      "loss": 1.2387,
+      "step": 309000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8753638310725467e-05,
+      "loss": 1.2377,
+      "step": 309500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.871931462463617e-05,
+      "loss": 1.2268,
+      "step": 310000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8684990938546875e-05,
+      "loss": 1.2371,
+      "step": 310500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8650667252457573e-05,
+      "loss": 1.2299,
+      "step": 311000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8616343566368282e-05,
+      "loss": 1.2268,
+      "step": 311500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.858201988027898e-05,
+      "loss": 1.2438,
+      "step": 312000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.854769619418969e-05,
+      "loss": 1.2413,
+      "step": 312500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8513372508100388e-05,
+      "loss": 1.2239,
+      "step": 313000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8479048822011097e-05,
+      "loss": 1.2381,
+      "step": 313500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8444725135921795e-05,
+      "loss": 1.2397,
+      "step": 314000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.84104014498325e-05,
+      "loss": 1.2396,
+      "step": 314500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8376077763743203e-05,
+      "loss": 1.2583,
+      "step": 315000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8341754077653908e-05,
+      "loss": 1.2382,
+      "step": 315500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.830743039156461e-05,
+      "loss": 1.2313,
+      "step": 316000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8273106705475316e-05,
+      "loss": 1.2273,
+      "step": 316500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8238783019386018e-05,
+      "loss": 1.2181,
+      "step": 317000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8204459333296723e-05,
+      "loss": 1.2368,
+      "step": 317500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8170135647207425e-05,
+      "loss": 1.241,
+      "step": 318000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.813581196111813e-05,
+      "loss": 1.2326,
+      "step": 318500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8101488275028832e-05,
+      "loss": 1.2294,
+      "step": 319000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8067164588939538e-05,
+      "loss": 1.2289,
+      "step": 319500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.803284090285024e-05,
+      "loss": 1.2148,
+      "step": 320000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7998517216760945e-05,
+      "loss": 1.2291,
+      "step": 320500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7964193530671644e-05,
+      "loss": 1.2252,
+      "step": 321000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7929869844582352e-05,
+      "loss": 1.2341,
+      "step": 321500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.789554615849305e-05,
+      "loss": 1.2222,
+      "step": 322000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.786122247240376e-05,
+      "loss": 1.2417,
+      "step": 322500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.782689878631446e-05,
+      "loss": 1.2423,
+      "step": 323000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7792575100225167e-05,
+      "loss": 1.2309,
+      "step": 323500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7758251414135866e-05,
+      "loss": 1.241,
+      "step": 324000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.772392772804657e-05,
+      "loss": 1.2541,
+      "step": 324500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7689604041957273e-05,
+      "loss": 1.2289,
+      "step": 325000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.765528035586798e-05,
+      "loss": 1.2203,
+      "step": 325500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.762095666977868e-05,
+      "loss": 1.2313,
+      "step": 326000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7586632983689386e-05,
+      "loss": 1.2185,
+      "step": 326500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7552309297600088e-05,
+      "loss": 1.234,
+      "step": 327000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7517985611510793e-05,
+      "loss": 1.2222,
+      "step": 327500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7483661925421495e-05,
+      "loss": 1.2157,
+      "step": 328000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.74493382393322e-05,
+      "loss": 1.2227,
+      "step": 328500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7415014553242903e-05,
+      "loss": 1.2353,
+      "step": 329000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7380690867153608e-05,
+      "loss": 1.2302,
+      "step": 329500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.734636718106431e-05,
+      "loss": 1.2297,
+      "step": 330000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7312043494975015e-05,
+      "loss": 1.2354,
+      "step": 330500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7277719808885714e-05,
+      "loss": 1.2441,
+      "step": 331000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7243396122796423e-05,
+      "loss": 1.2172,
+      "step": 331500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.720907243670712e-05,
+      "loss": 1.2282,
+      "step": 332000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.717474875061783e-05,
+      "loss": 1.2104,
+      "step": 332500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.714042506452853e-05,
+      "loss": 1.2259,
+      "step": 333000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7106101378439238e-05,
+      "loss": 1.2183,
+      "step": 333500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7071777692349936e-05,
+      "loss": 1.2079,
+      "step": 334000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.703745400626064e-05,
+      "loss": 1.2363,
+      "step": 334500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7003130320171343e-05,
+      "loss": 1.2242,
+      "step": 335000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.696880663408205e-05,
+      "loss": 1.2268,
+      "step": 335500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.693448294799275e-05,
+      "loss": 1.2199,
+      "step": 336000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6900159261903456e-05,
+      "loss": 1.2092,
+      "step": 336500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6865835575814158e-05,
+      "loss": 1.2225,
+      "step": 337000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6831511889724864e-05,
+      "loss": 1.2102,
+      "step": 337500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6797188203635566e-05,
+      "loss": 1.2093,
+      "step": 338000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.676286451754627e-05,
+      "loss": 1.2155,
+      "step": 338500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6728540831456973e-05,
+      "loss": 1.2225,
+      "step": 339000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.669421714536768e-05,
+      "loss": 1.2049,
+      "step": 339500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.665989345927838e-05,
+      "loss": 1.2069,
+      "step": 340000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6625569773189086e-05,
+      "loss": 1.222,
+      "step": 340500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6591246087099784e-05,
+      "loss": 1.204,
+      "step": 341000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6556922401010493e-05,
+      "loss": 1.2178,
+      "step": 341500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6522598714921192e-05,
+      "loss": 1.2127,
+      "step": 342000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.64882750288319e-05,
+      "loss": 1.2064,
+      "step": 342500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.64539513427426e-05,
+      "loss": 1.2185,
+      "step": 343000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6419627656653308e-05,
+      "loss": 1.2013,
+      "step": 343500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6385303970564006e-05,
+      "loss": 1.2292,
+      "step": 344000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6350980284474712e-05,
+      "loss": 1.2112,
+      "step": 344500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6316656598385414e-05,
+      "loss": 1.2157,
+      "step": 345000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.628233291229612e-05,
+      "loss": 1.2207,
+      "step": 345500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.624800922620682e-05,
+      "loss": 1.2022,
+      "step": 346000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6213685540117527e-05,
+      "loss": 1.2079,
+      "step": 346500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.617936185402823e-05,
+      "loss": 1.2268,
+      "step": 347000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6145038167938934e-05,
+      "loss": 1.2112,
+      "step": 347500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6110714481849636e-05,
+      "loss": 1.2163,
+      "step": 348000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.607639079576034e-05,
+      "loss": 1.2091,
+      "step": 348500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6042067109671043e-05,
+      "loss": 1.2165,
+      "step": 349000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.600774342358175e-05,
+      "loss": 1.2221,
+      "step": 349500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.597341973749245e-05,
+      "loss": 1.2155,
+      "step": 350000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5939096051403156e-05,
+      "loss": 1.2141,
+      "step": 350500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5904772365313855e-05,
+      "loss": 1.2207,
+      "step": 351000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5870448679224563e-05,
+      "loss": 1.2194,
+      "step": 351500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5836124993135262e-05,
+      "loss": 1.2293,
+      "step": 352000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.580180130704597e-05,
+      "loss": 1.2218,
+      "step": 352500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.576747762095667e-05,
+      "loss": 1.2016,
+      "step": 353000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5733153934867378e-05,
+      "loss": 1.2126,
+      "step": 353500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5698830248778077e-05,
+      "loss": 1.1931,
+      "step": 354000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5664506562688782e-05,
+      "loss": 1.2083,
+      "step": 354500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5630182876599484e-05,
+      "loss": 1.2098,
+      "step": 355000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.559585919051019e-05,
+      "loss": 1.2051,
+      "step": 355500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.556153550442089e-05,
+      "loss": 1.2199,
+      "step": 356000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5527211818331597e-05,
+      "loss": 1.2235,
+      "step": 356500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.54928881322423e-05,
+      "loss": 1.2155,
+      "step": 357000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5458564446153004e-05,
+      "loss": 1.2032,
+      "step": 357500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5424240760063706e-05,
+      "loss": 1.204,
+      "step": 358000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.538991707397441e-05,
+      "loss": 1.2113,
+      "step": 358500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5355593387885114e-05,
+      "loss": 1.213,
+      "step": 359000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.532126970179582e-05,
+      "loss": 1.21,
+      "step": 359500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.528694601570652e-05,
+      "loss": 1.2087,
+      "step": 360000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5252622329617226e-05,
+      "loss": 1.2143,
+      "step": 360500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.521829864352793e-05,
+      "loss": 1.1881,
+      "step": 361000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5183974957438634e-05,
+      "loss": 1.2134,
+      "step": 361500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5149651271349332e-05,
+      "loss": 1.2089,
+      "step": 362000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.511532758526004e-05,
+      "loss": 1.201,
+      "step": 362500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.508100389917074e-05,
+      "loss": 1.2026,
+      "step": 363000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.504668021308145e-05,
+      "loss": 1.1987,
+      "step": 363500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5012356526992147e-05,
+      "loss": 1.1874,
+      "step": 364000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4978032840902853e-05,
+      "loss": 1.2088,
+      "step": 364500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4943709154813554e-05,
+      "loss": 1.1979,
+      "step": 365000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.490938546872426e-05,
+      "loss": 1.2023,
+      "step": 365500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4875061782634962e-05,
+      "loss": 1.203,
+      "step": 366000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4840738096545664e-05,
+      "loss": 1.2179,
+      "step": 366500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.480641441045637e-05,
+      "loss": 1.1923,
+      "step": 367000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.477209072436707e-05,
+      "loss": 1.2023,
+      "step": 367500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4737767038277777e-05,
+      "loss": 1.205,
+      "step": 368000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.470344335218848e-05,
+      "loss": 1.2018,
+      "step": 368500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4669119666099184e-05,
+      "loss": 1.1879,
+      "step": 369000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4634795980009886e-05,
+      "loss": 1.1996,
+      "step": 369500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.460047229392059e-05,
+      "loss": 1.189,
+      "step": 370000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4566148607831293e-05,
+      "loss": 1.2067,
+      "step": 370500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4531824921742e-05,
+      "loss": 1.2127,
+      "step": 371000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44975012356527e-05,
+      "loss": 1.1895,
+      "step": 371500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4463177549563403e-05,
+      "loss": 1.2007,
+      "step": 372000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4428853863474108e-05,
+      "loss": 1.2043,
+      "step": 372500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.439453017738481e-05,
+      "loss": 1.1933,
+      "step": 373000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4360206491295515e-05,
+      "loss": 1.206,
+      "step": 373500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4325882805206217e-05,
+      "loss": 1.2027,
+      "step": 374000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4291559119116923e-05,
+      "loss": 1.1793,
+      "step": 374500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4257235433027625e-05,
+      "loss": 1.2073,
+      "step": 375000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.422291174693833e-05,
+      "loss": 1.1957,
+      "step": 375500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4188588060849032e-05,
+      "loss": 1.1937,
+      "step": 376000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4154264374759734e-05,
+      "loss": 1.1991,
+      "step": 376500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.411994068867044e-05,
+      "loss": 1.2111,
+      "step": 377000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.408561700258114e-05,
+      "loss": 1.1993,
+      "step": 377500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4051293316491847e-05,
+      "loss": 1.1972,
+      "step": 378000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.401696963040255e-05,
+      "loss": 1.1997,
+      "step": 378500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3982645944313254e-05,
+      "loss": 1.1913,
+      "step": 379000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3948322258223956e-05,
+      "loss": 1.1851,
+      "step": 379500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3913998572134662e-05,
+      "loss": 1.189,
+      "step": 380000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3879674886045364e-05,
+      "loss": 1.1992,
+      "step": 380500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.384535119995607e-05,
+      "loss": 1.1901,
+      "step": 381000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.381102751386677e-05,
+      "loss": 1.2061,
+      "step": 381500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3776703827777473e-05,
+      "loss": 1.1977,
+      "step": 382000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.374238014168818e-05,
+      "loss": 1.1913,
+      "step": 382500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.370805645559888e-05,
+      "loss": 1.1969,
+      "step": 383000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3673732769509586e-05,
+      "loss": 1.1941,
+      "step": 383500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3639409083420288e-05,
+      "loss": 1.2025,
+      "step": 384000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3605085397330993e-05,
+      "loss": 1.1943,
+      "step": 384500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3570761711241695e-05,
+      "loss": 1.1935,
+      "step": 385000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35364380251524e-05,
+      "loss": 1.1848,
+      "step": 385500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3502114339063103e-05,
+      "loss": 1.1874,
+      "step": 386000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3467790652973805e-05,
+      "loss": 1.1883,
+      "step": 386500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.343346696688451e-05,
+      "loss": 1.2087,
+      "step": 387000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3399143280795212e-05,
+      "loss": 1.1908,
+      "step": 387500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3364819594705917e-05,
+      "loss": 1.189,
+      "step": 388000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.333049590861662e-05,
+      "loss": 1.1851,
+      "step": 388500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3296172222527325e-05,
+      "loss": 1.1713,
+      "step": 389000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3261848536438027e-05,
+      "loss": 1.1979,
+      "step": 389500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3227524850348732e-05,
+      "loss": 1.1826,
+      "step": 390000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3193201164259434e-05,
+      "loss": 1.1974,
+      "step": 390500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.315887747817014e-05,
+      "loss": 1.2038,
+      "step": 391000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.312455379208084e-05,
+      "loss": 1.1846,
+      "step": 391500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3090230105991543e-05,
+      "loss": 1.1926,
+      "step": 392000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.305590641990225e-05,
+      "loss": 1.1905,
+      "step": 392500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.302158273381295e-05,
+      "loss": 1.1834,
+      "step": 393000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2987259047723656e-05,
+      "loss": 1.1923,
+      "step": 393500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2952935361634358e-05,
+      "loss": 1.1805,
+      "step": 394000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2918611675545064e-05,
+      "loss": 1.1868,
+      "step": 394500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2884287989455766e-05,
+      "loss": 1.1865,
+      "step": 395000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.284996430336647e-05,
+      "loss": 1.1906,
+      "step": 395500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2815640617277173e-05,
+      "loss": 1.2035,
+      "step": 396000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2781316931187878e-05,
+      "loss": 1.1846,
+      "step": 396500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.274699324509858e-05,
+      "loss": 1.1883,
+      "step": 397000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2712669559009282e-05,
+      "loss": 1.1768,
+      "step": 397500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2678345872919984e-05,
+      "loss": 1.1921,
+      "step": 398000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2644022186830686e-05,
+      "loss": 1.191,
+      "step": 398500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.260969850074139e-05,
+      "loss": 1.1961,
+      "step": 399000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2575374814652094e-05,
+      "loss": 1.1774,
+      "step": 399500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25410511285628e-05,
+      "loss": 1.1892,
+      "step": 400000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25067274424735e-05,
+      "loss": 1.1891,
+      "step": 400500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2472403756384206e-05,
+      "loss": 1.1744,
+      "step": 401000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.243808007029491e-05,
+      "loss": 1.1823,
+      "step": 401500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2403756384205614e-05,
+      "loss": 1.169,
+      "step": 402000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2369432698116316e-05,
+      "loss": 1.1818,
+      "step": 402500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.233510901202702e-05,
+      "loss": 1.1884,
+      "step": 403000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2300785325937723e-05,
+      "loss": 1.178,
+      "step": 403500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2266461639848425e-05,
+      "loss": 1.1778,
+      "step": 404000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.223213795375913e-05,
+      "loss": 1.1832,
+      "step": 404500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2197814267669832e-05,
+      "loss": 1.1826,
+      "step": 405000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2163490581580538e-05,
+      "loss": 1.1814,
+      "step": 405500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.212916689549124e-05,
+      "loss": 1.1768,
+      "step": 406000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2094843209401945e-05,
+      "loss": 1.1773,
+      "step": 406500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2060519523312647e-05,
+      "loss": 1.1911,
+      "step": 407000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2026195837223353e-05,
+      "loss": 1.1977,
+      "step": 407500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1991872151134055e-05,
+      "loss": 1.1783,
+      "step": 408000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1957548465044757e-05,
+      "loss": 1.1862,
+      "step": 408500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1923224778955462e-05,
+      "loss": 1.1748,
+      "step": 409000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1888901092866164e-05,
+      "loss": 1.1942,
+      "step": 409500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.185457740677687e-05,
+      "loss": 1.1826,
+      "step": 410000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.182025372068757e-05,
+      "loss": 1.1596,
+      "step": 410500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1785930034598277e-05,
+      "loss": 1.1854,
+      "step": 411000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.175160634850898e-05,
+      "loss": 1.1754,
+      "step": 411500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1717282662419684e-05,
+      "loss": 1.1721,
+      "step": 412000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1682958976330386e-05,
+      "loss": 1.1821,
+      "step": 412500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.164863529024109e-05,
+      "loss": 1.1721,
+      "step": 413000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1614311604151793e-05,
+      "loss": 1.1879,
+      "step": 413500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1579987918062495e-05,
+      "loss": 1.1748,
+      "step": 414000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.15456642319732e-05,
+      "loss": 1.1784,
+      "step": 414500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1511340545883903e-05,
+      "loss": 1.1718,
+      "step": 415000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1477016859794608e-05,
+      "loss": 1.1937,
+      "step": 415500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.144269317370531e-05,
+      "loss": 1.17,
+      "step": 416000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1408369487616016e-05,
+      "loss": 1.1795,
+      "step": 416500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1374045801526718e-05,
+      "loss": 1.1762,
+      "step": 417000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1339722115437423e-05,
+      "loss": 1.1813,
+      "step": 417500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1305398429348125e-05,
+      "loss": 1.1713,
+      "step": 418000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.127107474325883e-05,
+      "loss": 1.1697,
+      "step": 418500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1236751057169532e-05,
+      "loss": 1.1785,
+      "step": 419000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1202427371080234e-05,
+      "loss": 1.1781,
+      "step": 419500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.116810368499094e-05,
+      "loss": 1.1829,
+      "step": 420000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.113377999890164e-05,
+      "loss": 1.175,
+      "step": 420500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1099456312812347e-05,
+      "loss": 1.1739,
+      "step": 421000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.106513262672305e-05,
+      "loss": 1.1672,
+      "step": 421500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1030808940633754e-05,
+      "loss": 1.174,
+      "step": 422000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0996485254544456e-05,
+      "loss": 1.1741,
+      "step": 422500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0962161568455162e-05,
+      "loss": 1.1891,
+      "step": 423000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0927837882365864e-05,
+      "loss": 1.1718,
+      "step": 423500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0893514196276566e-05,
+      "loss": 1.176,
+      "step": 424000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.085919051018727e-05,
+      "loss": 1.169,
+      "step": 424500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0824866824097973e-05,
+      "loss": 1.1837,
+      "step": 425000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.079054313800868e-05,
+      "loss": 1.1749,
+      "step": 425500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.075621945191938e-05,
+      "loss": 1.161,
+      "step": 426000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0721895765830086e-05,
+      "loss": 1.1659,
+      "step": 426500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0687572079740788e-05,
+      "loss": 1.1625,
+      "step": 427000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0653248393651493e-05,
+      "loss": 1.192,
+      "step": 427500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0618924707562195e-05,
+      "loss": 1.1736,
+      "step": 428000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.05846010214729e-05,
+      "loss": 1.1756,
+      "step": 428500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0550277335383603e-05,
+      "loss": 1.1623,
+      "step": 429000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0515953649294305e-05,
+      "loss": 1.1548,
+      "step": 429500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.048162996320501e-05,
+      "loss": 1.1696,
+      "step": 430000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0447306277115712e-05,
+      "loss": 1.1738,
+      "step": 430500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0412982591026417e-05,
+      "loss": 1.1676,
+      "step": 431000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.037865890493712e-05,
+      "loss": 1.1667,
+      "step": 431500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0344335218847825e-05,
+      "loss": 1.1706,
+      "step": 432000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0310011532758527e-05,
+      "loss": 1.1759,
+      "step": 432500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0275687846669232e-05,
+      "loss": 1.1752,
+      "step": 433000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0241364160579934e-05,
+      "loss": 1.1697,
+      "step": 433500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0207040474490636e-05,
+      "loss": 1.1663,
+      "step": 434000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.017271678840134e-05,
+      "loss": 1.1607,
+      "step": 434500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0138393102312043e-05,
+      "loss": 1.1808,
+      "step": 435000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.010406941622275e-05,
+      "loss": 1.1732,
+      "step": 435500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.006974573013345e-05,
+      "loss": 1.1654,
+      "step": 436000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0035422044044156e-05,
+      "loss": 1.1624,
+      "step": 436500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0001098357954858e-05,
+      "loss": 1.1681,
+      "step": 437000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9966774671865564e-05,
+      "loss": 1.1435,
+      "step": 437500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9932450985776266e-05,
+      "loss": 1.1603,
+      "step": 438000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.989812729968697e-05,
+      "loss": 1.1743,
+      "step": 438500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9863803613597673e-05,
+      "loss": 1.1566,
+      "step": 439000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9829479927508375e-05,
+      "loss": 1.1655,
+      "step": 439500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979515624141908e-05,
+      "loss": 1.1692,
+      "step": 440000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9760832555329782e-05,
+      "loss": 1.1744,
+      "step": 440500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9726508869240488e-05,
+      "loss": 1.1791,
+      "step": 441000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.969218518315119e-05,
+      "loss": 1.1512,
+      "step": 441500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9657861497061895e-05,
+      "loss": 1.1546,
+      "step": 442000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9623537810972597e-05,
+      "loss": 1.1653,
+      "step": 442500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9589214124883302e-05,
+      "loss": 1.1763,
+      "step": 443000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9554890438794004e-05,
+      "loss": 1.1754,
+      "step": 443500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.952056675270471e-05,
+      "loss": 1.1755,
+      "step": 444000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9486243066615412e-05,
+      "loss": 1.1707,
+      "step": 444500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9451919380526114e-05,
+      "loss": 1.1642,
+      "step": 445000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.941759569443682e-05,
+      "loss": 1.149,
+      "step": 445500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.938327200834752e-05,
+      "loss": 1.1591,
+      "step": 446000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9348948322258227e-05,
+      "loss": 1.1637,
+      "step": 446500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.931462463616893e-05,
+      "loss": 1.1537,
+      "step": 447000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9280300950079634e-05,
+      "loss": 1.164,
+      "step": 447500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9245977263990336e-05,
+      "loss": 1.1678,
+      "step": 448000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.921165357790104e-05,
+      "loss": 1.1687,
+      "step": 448500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9177329891811743e-05,
+      "loss": 1.1532,
+      "step": 449000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9143006205722445e-05,
+      "loss": 1.1593,
+      "step": 449500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.910868251963315e-05,
+      "loss": 1.169,
+      "step": 450000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9074358833543853e-05,
+      "loss": 1.1605,
+      "step": 450500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9040035147454558e-05,
+      "loss": 1.1532,
+      "step": 451000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.900571146136526e-05,
+      "loss": 1.1509,
+      "step": 451500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8971387775275965e-05,
+      "loss": 1.1675,
+      "step": 452000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8937064089186667e-05,
+      "loss": 1.1575,
+      "step": 452500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8902740403097373e-05,
+      "loss": 1.1534,
+      "step": 453000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8868416717008075e-05,
+      "loss": 1.1656,
+      "step": 453500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.883409303091878e-05,
+      "loss": 1.16,
+      "step": 454000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8799769344829482e-05,
+      "loss": 1.1664,
+      "step": 454500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8765445658740184e-05,
+      "loss": 1.1549,
+      "step": 455000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.873112197265089e-05,
+      "loss": 1.1578,
+      "step": 455500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.869679828656159e-05,
+      "loss": 1.1543,
+      "step": 456000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8662474600472297e-05,
+      "loss": 1.1582,
+      "step": 456500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8628150914383e-05,
+      "loss": 1.1512,
+      "step": 457000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8593827228293704e-05,
+      "loss": 1.1682,
+      "step": 457500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8559503542204406e-05,
+      "loss": 1.1641,
+      "step": 458000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8525179856115108e-05,
+      "loss": 1.1379,
+      "step": 458500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.849085617002581e-05,
+      "loss": 1.1676,
+      "step": 459000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8456532483936516e-05,
+      "loss": 1.1628,
+      "step": 459500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8422208797847218e-05,
+      "loss": 1.1624,
+      "step": 460000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8387885111757923e-05,
+      "loss": 1.1337,
+      "step": 460500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353561425668625e-05,
+      "loss": 1.1584,
+      "step": 461000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319237739579327e-05,
+      "loss": 1.1721,
+      "step": 461500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8284914053490032e-05,
+      "loss": 1.1605,
+      "step": 462000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8250590367400734e-05,
+      "loss": 1.1683,
+      "step": 462500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.821626668131144e-05,
+      "loss": 1.1539,
+      "step": 463000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8181942995222142e-05,
+      "loss": 1.1537,
+      "step": 463500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8147619309132847e-05,
+      "loss": 1.1487,
+      "step": 464000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.811329562304355e-05,
+      "loss": 1.1566,
+      "step": 464500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8078971936954254e-05,
+      "loss": 1.1598,
+      "step": 465000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8044648250864956e-05,
+      "loss": 1.1632,
+      "step": 465500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8010324564775662e-05,
+      "loss": 1.1617,
+      "step": 466000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7976000878686364e-05,
+      "loss": 1.1549,
+      "step": 466500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7941677192597066e-05,
+      "loss": 1.1585,
+      "step": 467000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.790735350650777e-05,
+      "loss": 1.1581,
+      "step": 467500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7873029820418473e-05,
+      "loss": 1.1453,
+      "step": 468000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.783870613432918e-05,
+      "loss": 1.1329,
+      "step": 468500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.780438244823988e-05,
+      "loss": 1.1644,
+      "step": 469000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7770058762150586e-05,
+      "loss": 1.16,
+      "step": 469500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7735735076061288e-05,
+      "loss": 1.1575,
+      "step": 470000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7701411389971993e-05,
+      "loss": 1.1594,
+      "step": 470500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7667087703882695e-05,
+      "loss": 1.1563,
+      "step": 471000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7632764017793397e-05,
+      "loss": 1.1566,
+      "step": 471500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7598440331704103e-05,
+      "loss": 1.151,
+      "step": 472000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7564116645614805e-05,
+      "loss": 1.158,
+      "step": 472500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.752979295952551e-05,
+      "loss": 1.1673,
+      "step": 473000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7495469273436212e-05,
+      "loss": 1.1419,
+      "step": 473500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7461145587346917e-05,
+      "loss": 1.1342,
+      "step": 474000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.742682190125762e-05,
+      "loss": 1.1691,
+      "step": 474500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7392498215168325e-05,
+      "loss": 1.174,
+      "step": 475000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7358174529079027e-05,
+      "loss": 1.1565,
+      "step": 475500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7323850842989732e-05,
+      "loss": 1.1523,
+      "step": 476000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7289527156900434e-05,
+      "loss": 1.1378,
+      "step": 476500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7255203470811136e-05,
+      "loss": 1.1516,
+      "step": 477000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.722087978472184e-05,
+      "loss": 1.1512,
+      "step": 477500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7186556098632544e-05,
+      "loss": 1.1571,
+      "step": 478000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.715223241254325e-05,
+      "loss": 1.1408,
+      "step": 478500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.711790872645395e-05,
+      "loss": 1.1404,
+      "step": 479000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7083585040364656e-05,
+      "loss": 1.1433,
+      "step": 479500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.704926135427536e-05,
+      "loss": 1.1466,
+      "step": 480000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7014937668186064e-05,
+      "loss": 1.1476,
+      "step": 480500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6980613982096766e-05,
+      "loss": 1.1366,
+      "step": 481000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6946290296007468e-05,
+      "loss": 1.1394,
+      "step": 481500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6911966609918173e-05,
+      "loss": 1.1466,
+      "step": 482000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6877642923828875e-05,
+      "loss": 1.1489,
+      "step": 482500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.684331923773958e-05,
+      "loss": 1.1527,
+      "step": 483000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6808995551650282e-05,
+      "loss": 1.13,
+      "step": 483500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6774671865560988e-05,
+      "loss": 1.1549,
+      "step": 484000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.674034817947169e-05,
+      "loss": 1.1246,
+      "step": 484500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6706024493382395e-05,
+      "loss": 1.1478,
+      "step": 485000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6671700807293097e-05,
+      "loss": 1.1496,
+      "step": 485500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6637377121203803e-05,
+      "loss": 1.1511,
+      "step": 486000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6603053435114505e-05,
+      "loss": 1.1252,
+      "step": 486500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6568729749025207e-05,
+      "loss": 1.1503,
+      "step": 487000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6534406062935912e-05,
+      "loss": 1.1525,
+      "step": 487500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6500082376846614e-05,
+      "loss": 1.1391,
+      "step": 488000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.646575869075732e-05,
+      "loss": 1.1636,
+      "step": 488500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.643143500466802e-05,
+      "loss": 1.1541,
+      "step": 489000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6397111318578727e-05,
+      "loss": 1.1253,
+      "step": 489500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.636278763248943e-05,
+      "loss": 1.1437,
+      "step": 490000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6328463946400134e-05,
+      "loss": 1.1327,
+      "step": 490500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6294140260310836e-05,
+      "loss": 1.1451,
+      "step": 491000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6259816574221538e-05,
+      "loss": 1.1586,
+      "step": 491500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6225492888132243e-05,
+      "loss": 1.1416,
+      "step": 492000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6191169202042945e-05,
+      "loss": 1.1438,
+      "step": 492500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.615684551595365e-05,
+      "loss": 1.1366,
+      "step": 493000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6122521829864353e-05,
+      "loss": 1.1383,
+      "step": 493500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6088198143775058e-05,
+      "loss": 1.1445,
+      "step": 494000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.605387445768576e-05,
+      "loss": 1.1624,
+      "step": 494500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6019550771596466e-05,
+      "loss": 1.1331,
+      "step": 495000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5985227085507168e-05,
+      "loss": 1.1451,
+      "step": 495500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5950903399417873e-05,
+      "loss": 1.1531,
+      "step": 496000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5916579713328575e-05,
+      "loss": 1.1537,
+      "step": 496500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5882256027239277e-05,
+      "loss": 1.1514,
+      "step": 497000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5847932341149982e-05,
+      "loss": 1.1564,
+      "step": 497500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5813608655060684e-05,
+      "loss": 1.1311,
+      "step": 498000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.577928496897139e-05,
+      "loss": 1.1517,
+      "step": 498500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.574496128288209e-05,
+      "loss": 1.1388,
+      "step": 499000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5710637596792797e-05,
+      "loss": 1.1404,
+      "step": 499500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56763139107035e-05,
+      "loss": 1.1275,
+      "step": 500000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5641990224614204e-05,
+      "loss": 1.1455,
+      "step": 500500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5607666538524906e-05,
+      "loss": 1.1472,
+      "step": 501000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5573342852435612e-05,
+      "loss": 1.1409,
+      "step": 501500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5539019166346314e-05,
+      "loss": 1.1419,
+      "step": 502000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5504695480257016e-05,
+      "loss": 1.1368,
+      "step": 502500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.547037179416772e-05,
+      "loss": 1.1421,
+      "step": 503000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5436048108078423e-05,
+      "loss": 1.1307,
+      "step": 503500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.540172442198913e-05,
+      "loss": 1.1503,
+      "step": 504000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.536740073589983e-05,
+      "loss": 1.1415,
+      "step": 504500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5333077049810536e-05,
+      "loss": 1.1348,
+      "step": 505000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5298753363721238e-05,
+      "loss": 1.1459,
+      "step": 505500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5264429677631943e-05,
+      "loss": 1.1477,
+      "step": 506000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5230105991542645e-05,
+      "loss": 1.1372,
+      "step": 506500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5195782305453349e-05,
+      "loss": 1.1253,
+      "step": 507000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5161458619364053e-05,
+      "loss": 1.1266,
+      "step": 507500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5127134933274756e-05,
+      "loss": 1.1288,
+      "step": 508000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.509281124718546e-05,
+      "loss": 1.1388,
+      "step": 508500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5058487561096164e-05,
+      "loss": 1.1431,
+      "step": 509000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5024163875006866e-05,
+      "loss": 1.1306,
+      "step": 509500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.498984018891757e-05,
+      "loss": 1.1233,
+      "step": 510000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4955516502828273e-05,
+      "loss": 1.1334,
+      "step": 510500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4921192816738977e-05,
+      "loss": 1.1422,
+      "step": 511000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.488686913064968e-05,
+      "loss": 1.1234,
+      "step": 511500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4852545444560384e-05,
+      "loss": 1.1309,
+      "step": 512000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4818221758471088e-05,
+      "loss": 1.1296,
+      "step": 512500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4783898072381791e-05,
+      "loss": 1.1503,
+      "step": 513000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4749574386292495e-05,
+      "loss": 1.1445,
+      "step": 513500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4715250700203199e-05,
+      "loss": 1.1172,
+      "step": 514000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.46809270141139e-05,
+      "loss": 1.1197,
+      "step": 514500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4646603328024604e-05,
+      "loss": 1.116,
+      "step": 515000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4612279641935308e-05,
+      "loss": 1.1234,
+      "step": 515500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4577955955846012e-05,
+      "loss": 1.1256,
+      "step": 516000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4543632269756716e-05,
+      "loss": 1.1435,
+      "step": 516500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.450930858366742e-05,
+      "loss": 1.1163,
+      "step": 517000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4474984897578123e-05,
+      "loss": 1.1273,
+      "step": 517500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4440661211488827e-05,
+      "loss": 1.1213,
+      "step": 518000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.440633752539953e-05,
+      "loss": 1.1277,
+      "step": 518500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4372013839310234e-05,
+      "loss": 1.1185,
+      "step": 519000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4337690153220934e-05,
+      "loss": 1.1358,
+      "step": 519500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4303366467131638e-05,
+      "loss": 1.1315,
+      "step": 520000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4269042781042342e-05,
+      "loss": 1.1348,
+      "step": 520500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4234719094953045e-05,
+      "loss": 1.1121,
+      "step": 521000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4200395408863747e-05,
+      "loss": 1.132,
+      "step": 521500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4166071722774451e-05,
+      "loss": 1.1265,
+      "step": 522000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4131748036685155e-05,
+      "loss": 1.1156,
+      "step": 522500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4097424350595858e-05,
+      "loss": 1.1258,
+      "step": 523000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4063100664506562e-05,
+      "loss": 1.118,
+      "step": 523500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4028776978417266e-05,
+      "loss": 1.121,
+      "step": 524000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.399445329232797e-05,
+      "loss": 1.1362,
+      "step": 524500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3960129606238673e-05,
+      "loss": 1.1463,
+      "step": 525000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3925805920149377e-05,
+      "loss": 1.1338,
+      "step": 525500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.389148223406008e-05,
+      "loss": 1.1241,
+      "step": 526000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3857158547970783e-05,
+      "loss": 1.1262,
+      "step": 526500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3822834861881486e-05,
+      "loss": 1.1376,
+      "step": 527000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.378851117579219e-05,
+      "loss": 1.1267,
+      "step": 527500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3754187489702894e-05,
+      "loss": 1.1305,
+      "step": 528000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3719863803613597e-05,
+      "loss": 1.1264,
+      "step": 528500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3685540117524301e-05,
+      "loss": 1.1392,
+      "step": 529000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3651216431435005e-05,
+      "loss": 1.1286,
+      "step": 529500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3616892745345708e-05,
+      "loss": 1.1257,
+      "step": 530000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3582569059256412e-05,
+      "loss": 1.1205,
+      "step": 530500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3548245373167116e-05,
+      "loss": 1.1129,
+      "step": 531000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3513921687077818e-05,
+      "loss": 1.1415,
+      "step": 531500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3479598000988521e-05,
+      "loss": 1.1416,
+      "step": 532000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3445274314899225e-05,
+      "loss": 1.1123,
+      "step": 532500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3410950628809929e-05,
+      "loss": 1.1168,
+      "step": 533000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3376626942720632e-05,
+      "loss": 1.1235,
+      "step": 533500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3342303256631336e-05,
+      "loss": 1.1275,
+      "step": 534000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.330797957054204e-05,
+      "loss": 1.1171,
+      "step": 534500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3273655884452743e-05,
+      "loss": 1.1186,
+      "step": 535000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3239332198363447e-05,
+      "loss": 1.1296,
+      "step": 535500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.320500851227415e-05,
+      "loss": 1.1191,
+      "step": 536000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3170684826184853e-05,
+      "loss": 1.1147,
+      "step": 536500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3136361140095557e-05,
+      "loss": 1.1247,
+      "step": 537000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.310203745400626e-05,
+      "loss": 1.1293,
+      "step": 537500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3067713767916964e-05,
+      "loss": 1.1188,
+      "step": 538000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3033390081827668e-05,
+      "loss": 1.1189,
+      "step": 538500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2999066395738371e-05,
+      "loss": 1.1277,
+      "step": 539000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2964742709649075e-05,
+      "loss": 1.1154,
+      "step": 539500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2930419023559779e-05,
+      "loss": 1.1186,
+      "step": 540000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2896095337470482e-05,
+      "loss": 1.1305,
+      "step": 540500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2861771651381186e-05,
+      "loss": 1.1133,
+      "step": 541000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.282744796529189e-05,
+      "loss": 1.1174,
+      "step": 541500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2793124279202592e-05,
+      "loss": 1.1097,
+      "step": 542000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.2758800593113295e-05,
+      "loss": 1.1199,
+      "step": 542500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2724476907023999e-05,
+      "loss": 1.1168,
+      "step": 543000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2690153220934703e-05,
+      "loss": 1.1189,
+      "step": 543500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2655829534845406e-05,
+      "loss": 1.1175,
+      "step": 544000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.262150584875611e-05,
+      "loss": 1.1258,
+      "step": 544500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2587182162666814e-05,
+      "loss": 1.1275,
+      "step": 545000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2552858476577517e-05,
+      "loss": 1.1183,
+      "step": 545500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2518534790488221e-05,
+      "loss": 1.1264,
+      "step": 546000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2484211104398925e-05,
+      "loss": 1.1107,
+      "step": 546500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2449887418309627e-05,
+      "loss": 1.1207,
+      "step": 547000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.241556373222033e-05,
+      "loss": 1.1197,
+      "step": 547500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2381240046131034e-05,
+      "loss": 1.1133,
+      "step": 548000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2346916360041738e-05,
+      "loss": 1.1144,
+      "step": 548500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2312592673952442e-05,
+      "loss": 1.1179,
+      "step": 549000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.2278268987863145e-05,
+      "loss": 1.103,
+      "step": 549500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2243945301773849e-05,
+      "loss": 1.1107,
+      "step": 550000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2209621615684553e-05,
+      "loss": 1.1112,
+      "step": 550500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2175297929595256e-05,
+      "loss": 1.1213,
+      "step": 551000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.214097424350596e-05,
+      "loss": 1.1145,
+      "step": 551500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2106650557416662e-05,
+      "loss": 1.1192,
+      "step": 552000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2072326871327366e-05,
+      "loss": 1.1066,
+      "step": 552500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.203800318523807e-05,
+      "loss": 1.1142,
+      "step": 553000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.2003679499148773e-05,
+      "loss": 1.1231,
+      "step": 553500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1969355813059477e-05,
+      "loss": 1.1043,
+      "step": 554000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.193503212697018e-05,
+      "loss": 1.1253,
+      "step": 554500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1900708440880884e-05,
+      "loss": 1.1159,
+      "step": 555000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1866384754791588e-05,
+      "loss": 1.1332,
+      "step": 555500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1832061068702292e-05,
+      "loss": 1.1203,
+      "step": 556000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1797737382612995e-05,
+      "loss": 1.1176,
+      "step": 556500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1763413696523697e-05,
+      "loss": 1.1288,
+      "step": 557000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1729090010434401e-05,
+      "loss": 1.1123,
+      "step": 557500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1694766324345105e-05,
+      "loss": 1.1144,
+      "step": 558000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1660442638255808e-05,
+      "loss": 1.1042,
+      "step": 558500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1626118952166512e-05,
+      "loss": 1.1046,
+      "step": 559000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1591795266077216e-05,
+      "loss": 1.1178,
+      "step": 559500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.155747157998792e-05,
+      "loss": 1.1204,
+      "step": 560000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1523147893898623e-05,
+      "loss": 1.122,
+      "step": 560500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1488824207809327e-05,
+      "loss": 1.1163,
+      "step": 561000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.145450052172003e-05,
+      "loss": 1.1089,
+      "step": 561500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1420176835630732e-05,
+      "loss": 1.0995,
+      "step": 562000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1385853149541436e-05,
+      "loss": 1.1071,
+      "step": 562500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.135152946345214e-05,
+      "loss": 1.1279,
+      "step": 563000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1317205777362843e-05,
+      "loss": 1.1133,
+      "step": 563500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1282882091273547e-05,
+      "loss": 1.1104,
+      "step": 564000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.124855840518425e-05,
+      "loss": 1.1127,
+      "step": 564500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1214234719094954e-05,
+      "loss": 1.1086,
+      "step": 565000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1179911033005656e-05,
+      "loss": 1.1049,
+      "step": 565500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.114558734691636e-05,
+      "loss": 1.1204,
+      "step": 566000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1111263660827064e-05,
+      "loss": 1.1175,
+      "step": 566500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1076939974737768e-05,
+      "loss": 1.1029,
+      "step": 567000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1042616288648471e-05,
+      "loss": 1.1157,
+      "step": 567500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1008292602559173e-05,
+      "loss": 1.1129,
+      "step": 568000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0973968916469877e-05,
+      "loss": 1.119,
+      "step": 568500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.093964523038058e-05,
+      "loss": 1.1126,
+      "step": 569000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0905321544291284e-05,
+      "loss": 1.1201,
+      "step": 569500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0870997858201988e-05,
+      "loss": 1.1253,
+      "step": 570000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0836674172112692e-05,
+      "loss": 1.1019,
+      "step": 570500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0802350486023395e-05,
+      "loss": 1.1002,
+      "step": 571000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.0768026799934099e-05,
+      "loss": 1.1029,
+      "step": 571500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0733703113844803e-05,
+      "loss": 1.1161,
+      "step": 572000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0699379427755506e-05,
+      "loss": 1.0901,
+      "step": 572500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0665055741666208e-05,
+      "loss": 1.1245,
+      "step": 573000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0630732055576912e-05,
+      "loss": 1.1055,
+      "step": 573500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0596408369487616e-05,
+      "loss": 1.1068,
+      "step": 574000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.056208468339832e-05,
+      "loss": 1.0998,
+      "step": 574500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0527760997309023e-05,
+      "loss": 1.0957,
+      "step": 575000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0493437311219727e-05,
+      "loss": 1.0896,
+      "step": 575500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.045911362513043e-05,
+      "loss": 1.0978,
+      "step": 576000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0424789939041134e-05,
+      "loss": 1.106,
+      "step": 576500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0390466252951838e-05,
+      "loss": 1.1152,
+      "step": 577000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0356142566862542e-05,
+      "loss": 1.0977,
+      "step": 577500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0321818880773244e-05,
+      "loss": 1.1115,
+      "step": 578000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0287495194683947e-05,
+      "loss": 1.0966,
+      "step": 578500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0253171508594651e-05,
+      "loss": 1.1201,
+      "step": 579000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0218847822505355e-05,
+      "loss": 1.1168,
+      "step": 579500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0184524136416058e-05,
+      "loss": 1.1141,
+      "step": 580000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0150200450326762e-05,
+      "loss": 1.1115,
+      "step": 580500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0115876764237466e-05,
+      "loss": 1.1158,
+      "step": 581000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.008155307814817e-05,
+      "loss": 1.1131,
+      "step": 581500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0047229392058873e-05,
+      "loss": 1.1007,
+      "step": 582000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0012905705969577e-05,
+      "loss": 1.1066,
+      "step": 582500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.97858201988028e-06,
+      "loss": 1.1094,
+      "step": 583000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.944258333790982e-06,
+      "loss": 1.0962,
+      "step": 583500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.909934647701686e-06,
+      "loss": 1.0962,
+      "step": 584000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.87561096161239e-06,
+      "loss": 1.1004,
+      "step": 584500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.841287275523093e-06,
+      "loss": 1.1165,
+      "step": 585000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.806963589433797e-06,
+      "loss": 1.1079,
+      "step": 585500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.7726399033445e-06,
+      "loss": 1.1008,
+      "step": 586000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.738316217255205e-06,
+      "loss": 1.0872,
+      "step": 586500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.703992531165908e-06,
+      "loss": 1.1128,
+      "step": 587000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.669668845076612e-06,
+      "loss": 1.1272,
+      "step": 587500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.635345158987316e-06,
+      "loss": 1.1073,
+      "step": 588000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.601021472898018e-06,
+      "loss": 1.0932,
+      "step": 588500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.566697786808721e-06,
+      "loss": 1.1003,
+      "step": 589000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.532374100719425e-06,
+      "loss": 1.1132,
+      "step": 589500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.498050414630129e-06,
+      "loss": 1.0932,
+      "step": 590000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.463726728540832e-06,
+      "loss": 1.1129,
+      "step": 590500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.429403042451536e-06,
+      "loss": 1.1182,
+      "step": 591000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.39507935636224e-06,
+      "loss": 1.1044,
+      "step": 591500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.360755670272943e-06,
+      "loss": 1.0989,
+      "step": 592000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.326431984183647e-06,
+      "loss": 1.1118,
+      "step": 592500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.29210829809435e-06,
+      "loss": 1.0983,
+      "step": 593000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.257784612005053e-06,
+      "loss": 1.1071,
+      "step": 593500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.223460925915756e-06,
+      "loss": 1.091,
+      "step": 594000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.18913723982646e-06,
+      "loss": 1.0953,
+      "step": 594500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.154813553737164e-06,
+      "loss": 1.1123,
+      "step": 595000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.120489867647867e-06,
+      "loss": 1.1069,
+      "step": 595500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.08616618155857e-06,
+      "loss": 1.0998,
+      "step": 596000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.051842495469273e-06,
+      "loss": 1.0986,
+      "step": 596500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.017518809379977e-06,
+      "loss": 1.1044,
+      "step": 597000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.98319512329068e-06,
+      "loss": 1.0957,
+      "step": 597500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.948871437201384e-06,
+      "loss": 1.0912,
+      "step": 598000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.914547751112088e-06,
+      "loss": 1.095,
+      "step": 598500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.880224065022792e-06,
+      "loss": 1.0936,
+      "step": 599000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.845900378933494e-06,
+      "loss": 1.0951,
+      "step": 599500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.811576692844197e-06,
+      "loss": 1.1066,
+      "step": 600000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.777253006754901e-06,
+      "loss": 1.0902,
+      "step": 600500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.742929320665605e-06,
+      "loss": 1.0947,
+      "step": 601000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.708605634576308e-06,
+      "loss": 1.0913,
+      "step": 601500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.674281948487012e-06,
+      "loss": 1.089,
+      "step": 602000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.639958262397716e-06,
+      "loss": 1.1062,
+      "step": 602500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.60563457630842e-06,
+      "loss": 1.1013,
+      "step": 603000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.571310890219123e-06,
+      "loss": 1.09,
+      "step": 603500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.536987204129827e-06,
+      "loss": 1.1014,
+      "step": 604000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.502663518040529e-06,
+      "loss": 1.1057,
+      "step": 604500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.468339831951232e-06,
+      "loss": 1.088,
+      "step": 605000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.434016145861936e-06,
+      "loss": 1.11,
+      "step": 605500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.39969245977264e-06,
+      "loss": 1.1013,
+      "step": 606000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.365368773683344e-06,
+      "loss": 1.0868,
+      "step": 606500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.331045087594047e-06,
+      "loss": 1.1111,
+      "step": 607000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.296721401504751e-06,
+      "loss": 1.1081,
+      "step": 607500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.262397715415455e-06,
+      "loss": 1.1064,
+      "step": 608000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.228074029326158e-06,
+      "loss": 1.0929,
+      "step": 608500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.193750343236862e-06,
+      "loss": 1.0806,
+      "step": 609000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.159426657147564e-06,
+      "loss": 1.0952,
+      "step": 609500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.125102971058268e-06,
+      "loss": 1.0859,
+      "step": 610000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.090779284968971e-06,
+      "loss": 1.1025,
+      "step": 610500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.056455598879675e-06,
+      "loss": 1.1,
+      "step": 611000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.022131912790379e-06,
+      "loss": 1.0921,
+      "step": 611500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.987808226701082e-06,
+      "loss": 1.1135,
+      "step": 612000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.953484540611786e-06,
+      "loss": 1.1048,
+      "step": 612500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.91916085452249e-06,
+      "loss": 1.0836,
+      "step": 613000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.884837168433193e-06,
+      "loss": 1.0704,
+      "step": 613500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.850513482343897e-06,
+      "loss": 1.0878,
+      "step": 614000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.816189796254599e-06,
+      "loss": 1.1043,
+      "step": 614500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.781866110165303e-06,
+      "loss": 1.0907,
+      "step": 615000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.747542424076006e-06,
+      "loss": 1.0865,
+      "step": 615500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.71321873798671e-06,
+      "loss": 1.1002,
+      "step": 616000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.678895051897414e-06,
+      "loss": 1.0805,
+      "step": 616500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.644571365808118e-06,
+      "loss": 1.0951,
+      "step": 617000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.610247679718821e-06,
+      "loss": 1.0911,
+      "step": 617500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.575923993629525e-06,
+      "loss": 1.1089,
+      "step": 618000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.541600307540228e-06,
+      "loss": 1.0927,
+      "step": 618500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.5072766214509314e-06,
+      "loss": 1.0991,
+      "step": 619000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.472952935361635e-06,
+      "loss": 1.086,
+      "step": 619500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.438629249272339e-06,
+      "loss": 1.0876,
+      "step": 620000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.4043055631830425e-06,
+      "loss": 1.0998,
+      "step": 620500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.369981877093745e-06,
+      "loss": 1.0976,
+      "step": 621000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.335658191004449e-06,
+      "loss": 1.0973,
+      "step": 621500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.301334504915153e-06,
+      "loss": 1.0889,
+      "step": 622000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.267010818825856e-06,
+      "loss": 1.1159,
+      "step": 622500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.23268713273656e-06,
+      "loss": 1.1006,
+      "step": 623000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.198363446647263e-06,
+      "loss": 1.1102,
+      "step": 623500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.164039760557967e-06,
+      "loss": 1.0785,
+      "step": 624000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.12971607446867e-06,
+      "loss": 1.0885,
+      "step": 624500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.095392388379374e-06,
+      "loss": 1.0783,
+      "step": 625000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.061068702290078e-06,
+      "loss": 1.0796,
+      "step": 625500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.0267450162007805e-06,
+      "loss": 1.0923,
+      "step": 626000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.992421330111483e-06,
+      "loss": 1.0983,
+      "step": 626500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.958097644022186e-06,
+      "loss": 1.0887,
+      "step": 627000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.92377395793289e-06,
+      "loss": 1.0974,
+      "step": 627500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.8894502718435936e-06,
+      "loss": 1.0868,
+      "step": 628000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.855126585754297e-06,
+      "loss": 1.0818,
+      "step": 628500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.820802899665001e-06,
+      "loss": 1.085,
+      "step": 629000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.786479213575704e-06,
+      "loss": 1.0858,
+      "step": 629500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.7521555274864075e-06,
+      "loss": 1.0874,
+      "step": 630000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.717831841397111e-06,
+      "loss": 1.0957,
+      "step": 630500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.683508155307815e-06,
+      "loss": 1.0764,
+      "step": 631000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.6491844692185185e-06,
+      "loss": 1.1055,
+      "step": 631500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.614860783129221e-06,
+      "loss": 1.0771,
+      "step": 632000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.580537097039925e-06,
+      "loss": 1.093,
+      "step": 632500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.546213410950629e-06,
+      "loss": 1.0797,
+      "step": 633000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.511889724861332e-06,
+      "loss": 1.0846,
+      "step": 633500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.477566038772036e-06,
+      "loss": 1.0957,
+      "step": 634000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.443242352682739e-06,
+      "loss": 1.0838,
+      "step": 634500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.408918666593443e-06,
+      "loss": 1.0865,
+      "step": 635000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.374594980504146e-06,
+      "loss": 1.0796,
+      "step": 635500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.34027129441485e-06,
+      "loss": 1.0867,
+      "step": 636000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.305947608325554e-06,
+      "loss": 1.0906,
+      "step": 636500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.2716239222362565e-06,
+      "loss": 1.0798,
+      "step": 637000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.23730023614696e-06,
+      "loss": 1.1051,
+      "step": 637500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.202976550057664e-06,
+      "loss": 1.0906,
+      "step": 638000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.168652863968368e-06,
+      "loss": 1.0716,
+      "step": 638500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.134329177879071e-06,
+      "loss": 1.0777,
+      "step": 639000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.100005491789774e-06,
+      "loss": 1.108,
+      "step": 639500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.065681805700478e-06,
+      "loss": 1.0964,
+      "step": 640000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.0313581196111815e-06,
+      "loss": 1.1027,
+      "step": 640500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.997034433521885e-06,
+      "loss": 1.0936,
+      "step": 641000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.962710747432589e-06,
+      "loss": 1.0964,
+      "step": 641500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.928387061343292e-06,
+      "loss": 1.0898,
+      "step": 642000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894063375253995e-06,
+      "loss": 1.0904,
+      "step": 642500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.859739689164699e-06,
+      "loss": 1.0861,
+      "step": 643000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.825416003075403e-06,
+      "loss": 1.1034,
+      "step": 643500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.7910923169861064e-06,
+      "loss": 1.0817,
+      "step": 644000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.756768630896809e-06,
+      "loss": 1.0789,
+      "step": 644500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.722444944807513e-06,
+      "loss": 1.085,
+      "step": 645000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.688121258718217e-06,
+      "loss": 1.088,
+      "step": 645500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.65379757262892e-06,
+      "loss": 1.083,
+      "step": 646000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.619473886539624e-06,
+      "loss": 1.0947,
+      "step": 646500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.585150200450327e-06,
+      "loss": 1.085,
+      "step": 647000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5508265143610305e-06,
+      "loss": 1.0759,
+      "step": 647500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.516502828271734e-06,
+      "loss": 1.0871,
+      "step": 648000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.482179142182438e-06,
+      "loss": 1.0826,
+      "step": 648500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.447855456093142e-06,
+      "loss": 1.0747,
+      "step": 649000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.4135317700038444e-06,
+      "loss": 1.0902,
+      "step": 649500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.379208083914547e-06,
+      "loss": 1.0796,
+      "step": 650000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.344884397825251e-06,
+      "loss": 1.0885,
+      "step": 650500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.310560711735955e-06,
+      "loss": 1.0806,
+      "step": 651000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.276237025646658e-06,
+      "loss": 1.0777,
+      "step": 651500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.241913339557362e-06,
+      "loss": 1.0792,
+      "step": 652000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.207589653468066e-06,
+      "loss": 1.0692,
+      "step": 652500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1732659673787686e-06,
+      "loss": 1.086,
+      "step": 653000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.138942281289472e-06,
+      "loss": 1.0826,
+      "step": 653500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104618595200176e-06,
+      "loss": 1.0807,
+      "step": 654000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07029490911088e-06,
+      "loss": 1.094,
+      "step": 654500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.035971223021583e-06,
+      "loss": 1.0893,
+      "step": 655000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001647536932286e-06,
+      "loss": 1.077,
+      "step": 655500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.96732385084299e-06,
+      "loss": 1.0764,
+      "step": 656000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9330001647536935e-06,
+      "loss": 1.083,
+      "step": 656500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898676478664397e-06,
+      "loss": 1.0704,
+      "step": 657000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864352792575101e-06,
+      "loss": 1.0803,
+      "step": 657500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.830029106485804e-06,
+      "loss": 1.0949,
+      "step": 658000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.795705420396507e-06,
+      "loss": 1.0691,
+      "step": 658500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.761381734307211e-06,
+      "loss": 1.067,
+      "step": 659000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.727058048217915e-06,
+      "loss": 1.0707,
+      "step": 659500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6927343621286185e-06,
+      "loss": 1.0976,
+      "step": 660000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.658410676039321e-06,
+      "loss": 1.0884,
+      "step": 660500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.624086989950025e-06,
+      "loss": 1.0792,
+      "step": 661000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.589763303860729e-06,
+      "loss": 1.0864,
+      "step": 661500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555439617771432e-06,
+      "loss": 1.0789,
+      "step": 662000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.521115931682136e-06,
+      "loss": 1.0789,
+      "step": 662500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.486792245592839e-06,
+      "loss": 1.0631,
+      "step": 663000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.452468559503543e-06,
+      "loss": 1.0762,
+      "step": 663500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.418144873414246e-06,
+      "loss": 1.0925,
+      "step": 664000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.38382118732495e-06,
+      "loss": 1.0835,
+      "step": 664500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.349497501235653e-06,
+      "loss": 1.0793,
+      "step": 665000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3151738151463565e-06,
+      "loss": 1.0744,
+      "step": 665500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.280850129057059e-06,
+      "loss": 1.0862,
+      "step": 666000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.246526442967763e-06,
+      "loss": 1.093,
+      "step": 666500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.212202756878467e-06,
+      "loss": 1.0831,
+      "step": 667000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.17787907078917e-06,
+      "loss": 1.0688,
+      "step": 667500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.143555384699874e-06,
+      "loss": 1.083,
+      "step": 668000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.109231698610577e-06,
+      "loss": 1.076,
+      "step": 668500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.074908012521281e-06,
+      "loss": 1.077,
+      "step": 669000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.040584326431984e-06,
+      "loss": 1.0744,
+      "step": 669500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.006260640342688e-06,
+      "loss": 1.0867,
+      "step": 670000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.971936954253392e-06,
+      "loss": 1.0703,
+      "step": 670500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9376132681640945e-06,
+      "loss": 1.0791,
+      "step": 671000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.903289582074798e-06,
+      "loss": 1.0697,
+      "step": 671500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.868965895985502e-06,
+      "loss": 1.0797,
+      "step": 672000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8346422098962055e-06,
+      "loss": 1.0853,
+      "step": 672500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.800318523806909e-06,
+      "loss": 1.0701,
+      "step": 673000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7659948377176125e-06,
+      "loss": 1.0709,
+      "step": 673500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.7316711516283158e-06,
+      "loss": 1.0844,
+      "step": 674000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6973474655390194e-06,
+      "loss": 1.0955,
+      "step": 674500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.663023779449723e-06,
+      "loss": 1.0801,
+      "step": 675000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6287000933604264e-06,
+      "loss": 1.0712,
+      "step": 675500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.59437640727113e-06,
+      "loss": 1.0934,
+      "step": 676000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5600527211818333e-06,
+      "loss": 1.0748,
+      "step": 676500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.525729035092537e-06,
+      "loss": 1.0807,
+      "step": 677000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4914053490032407e-06,
+      "loss": 1.0654,
+      "step": 677500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457081662913944e-06,
+      "loss": 1.0735,
+      "step": 678000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4227579768246477e-06,
+      "loss": 1.0799,
+      "step": 678500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.388434290735351e-06,
+      "loss": 1.0725,
+      "step": 679000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3541106046460546e-06,
+      "loss": 1.0753,
+      "step": 679500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3197869185567574e-06,
+      "loss": 1.0753,
+      "step": 680000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.285463232467461e-06,
+      "loss": 1.0686,
+      "step": 680500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2511395463781644e-06,
+      "loss": 1.0728,
+      "step": 681000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.216815860288868e-06,
+      "loss": 1.0814,
+      "step": 681500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1824921741995713e-06,
+      "loss": 1.0723,
+      "step": 682000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.148168488110275e-06,
+      "loss": 1.0686,
+      "step": 682500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1138448020209787e-06,
+      "loss": 1.0765,
+      "step": 683000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0795211159316824e-06,
+      "loss": 1.0784,
+      "step": 683500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0451974298423857e-06,
+      "loss": 1.0658,
+      "step": 684000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.010873743753089e-06,
+      "loss": 1.0706,
+      "step": 684500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9765500576637926e-06,
+      "loss": 1.064,
+      "step": 685000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9422263715744963e-06,
+      "loss": 1.0607,
+      "step": 685500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9079026854851996e-06,
+      "loss": 1.0754,
+      "step": 686000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8735789993959033e-06,
+      "loss": 1.0703,
+      "step": 686500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.839255313306607e-06,
+      "loss": 1.075,
+      "step": 687000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.80493162721731e-06,
+      "loss": 1.0853,
+      "step": 687500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.770607941128014e-06,
+      "loss": 1.0648,
+      "step": 688000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.736284255038717e-06,
+      "loss": 1.0669,
+      "step": 688500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.701960568949421e-06,
+      "loss": 1.0867,
+      "step": 689000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6676368828601245e-06,
+      "loss": 1.0738,
+      "step": 689500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6333131967708278e-06,
+      "loss": 1.0695,
+      "step": 690000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5989895106815315e-06,
+      "loss": 1.0743,
+      "step": 690500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5646658245922347e-06,
+      "loss": 1.0844,
+      "step": 691000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.530342138502938e-06,
+      "loss": 1.0666,
+      "step": 691500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4960184524136417e-06,
+      "loss": 1.0844,
+      "step": 692000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.461694766324345e-06,
+      "loss": 1.0772,
+      "step": 692500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4273710802350486e-06,
+      "loss": 1.0795,
+      "step": 693000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3930473941457523e-06,
+      "loss": 1.083,
+      "step": 693500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3587237080564556e-06,
+      "loss": 1.0688,
+      "step": 694000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.3244000219671593e-06,
+      "loss": 1.0663,
+      "step": 694500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2900763358778625e-06,
+      "loss": 1.0721,
+      "step": 695000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2557526497885662e-06,
+      "loss": 1.0708,
+      "step": 695500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.22142896369927e-06,
+      "loss": 1.0769,
+      "step": 696000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.187105277609973e-06,
+      "loss": 1.0713,
+      "step": 696500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.152781591520677e-06,
+      "loss": 1.0719,
+      "step": 697000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.11845790543138e-06,
+      "loss": 1.0891,
+      "step": 697500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.084134219342084e-06,
+      "loss": 1.0772,
+      "step": 698000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0498105332527875e-06,
+      "loss": 1.0737,
+      "step": 698500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0154868471634903e-06,
+      "loss": 1.073,
+      "step": 699000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.981163161074194e-06,
+      "loss": 1.0852,
+      "step": 699500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9468394749848977e-06,
+      "loss": 1.0722,
+      "step": 700000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.912515788895601e-06,
+      "loss": 1.0753,
+      "step": 700500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8781921028063047e-06,
+      "loss": 1.0826,
+      "step": 701000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8438684167170081e-06,
+      "loss": 1.0649,
+      "step": 701500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8095447306277116e-06,
+      "loss": 1.0634,
+      "step": 702000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.775221044538415e-06,
+      "loss": 1.0606,
+      "step": 702500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7408973584491188e-06,
+      "loss": 1.0734,
+      "step": 703000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7065736723598222e-06,
+      "loss": 1.0655,
+      "step": 703500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6722499862705257e-06,
+      "loss": 1.0617,
+      "step": 704000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6379263001812292e-06,
+      "loss": 1.0674,
+      "step": 704500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6036026140919327e-06,
+      "loss": 1.0725,
+      "step": 705000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5692789280026363e-06,
+      "loss": 1.0747,
+      "step": 705500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5349552419133396e-06,
+      "loss": 1.0743,
+      "step": 706000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.500631555824043e-06,
+      "loss": 1.0667,
+      "step": 706500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4663078697347466e-06,
+      "loss": 1.0732,
+      "step": 707000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4319841836454502e-06,
+      "loss": 1.0785,
+      "step": 707500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3976604975561537e-06,
+      "loss": 1.0849,
+      "step": 708000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.363336811466857e-06,
+      "loss": 1.068,
+      "step": 708500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3290131253775605e-06,
+      "loss": 1.0581,
+      "step": 709000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2946894392882641e-06,
+      "loss": 1.0778,
+      "step": 709500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2603657531989676e-06,
+      "loss": 1.0694,
+      "step": 710000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.226042067109671e-06,
+      "loss": 1.0748,
+      "step": 710500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1917183810203746e-06,
+      "loss": 1.0765,
+      "step": 711000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.157394694931078e-06,
+      "loss": 1.0645,
+      "step": 711500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1230710088417817e-06,
+      "loss": 1.0692,
+      "step": 712000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.088747322752485e-06,
+      "loss": 1.0628,
+      "step": 712500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0544236366631885e-06,
+      "loss": 1.0544,
+      "step": 713000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0200999505738922e-06,
+      "loss": 1.068,
+      "step": 713500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.857762644845956e-07,
+      "loss": 1.0732,
+      "step": 714000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.514525783952991e-07,
+      "loss": 1.0678,
+      "step": 714500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.171288923060026e-07,
+      "loss": 1.0572,
+      "step": 715000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.828052062167062e-07,
+      "loss": 1.0625,
+      "step": 715500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.484815201274096e-07,
+      "loss": 1.0809,
+      "step": 716000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.14157834038113e-07,
+      "loss": 1.0675,
+      "step": 716500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.798341479488166e-07,
+      "loss": 1.0576,
+      "step": 717000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.455104618595201e-07,
+      "loss": 1.0636,
+      "step": 717500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.111867757702235e-07,
+      "loss": 1.0804,
+      "step": 718000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.768630896809271e-07,
+      "loss": 1.0678,
+      "step": 718500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.425394035916305e-07,
+      "loss": 1.0545,
+      "step": 719000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.082157175023341e-07,
+      "loss": 1.0657,
+      "step": 719500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.738920314130375e-07,
+      "loss": 1.0486,
+      "step": 720000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.395683453237411e-07,
+      "loss": 1.063,
+      "step": 720500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.052446592344445e-07,
+      "loss": 1.0733,
+      "step": 721000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.70920973145148e-07,
+      "loss": 1.0695,
+      "step": 721500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3659728705585154e-07,
+      "loss": 1.0728,
+      "step": 722000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0227360096655506e-07,
+      "loss": 1.0687,
+      "step": 722500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6794991487725854e-07,
+      "loss": 1.0717,
+      "step": 723000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.33626228787962e-07,
+      "loss": 1.0643,
+      "step": 723500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9930254269866554e-07,
+      "loss": 1.0738,
+      "step": 724000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.64978856609369e-07,
+      "loss": 1.0722,
+      "step": 724500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3065517052007252e-07,
+      "loss": 1.0768,
+      "step": 725000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.96331484430776e-07,
+      "loss": 1.0689,
+      "step": 725500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.620077983414795e-07,
+      "loss": 1.0709,
+      "step": 726000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.27684112252183e-07,
+      "loss": 1.0672,
+      "step": 726500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.336042616288648e-08,
+      "loss": 1.0662,
+      "step": 727000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.903674007358998e-08,
+      "loss": 1.0778,
+      "step": 727500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4713053984293483e-08,
+      "loss": 1.0693,
+      "step": 728000
+    }
+  ],
+  "max_steps": 728360,
+  "num_train_epochs": 1,
+  "total_flos": 3.0658096594944e+18,
+  "trial_name": null,
+  "trial_params": null
+}