diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,16816 +1,18136 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9994610049580405,
-  "global_step": 280000,
+  "epoch": 0.9991860934470597,
+  "global_step": 302000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 4.9982152482054324e-05,
-      "loss": 3.2655,
+      "learning_rate": 4.9983430266275624e-05,
+      "loss": 3.1363,
       "step": 100
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.9964304964108645e-05,
-      "loss": 3.2142,
+      "learning_rate": 4.9966860532551245e-05,
+      "loss": 3.1155,
       "step": 200
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.9946457446162966e-05,
-      "loss": 3.1805,
+      "learning_rate": 4.995029079882687e-05,
+      "loss": 3.1139,
       "step": 300
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.992860992821729e-05,
-      "loss": 3.1781,
+      "learning_rate": 4.993372106510248e-05,
+      "loss": 3.124,
       "step": 400
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.991076241027161e-05,
-      "loss": 3.1664,
+      "learning_rate": 4.991715133137811e-05,
+      "loss": 3.1111,
       "step": 500
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.989291489232592e-05,
-      "loss": 3.158,
+      "learning_rate": 4.990058159765373e-05,
+      "loss": 3.1051,
       "step": 600
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.9875067374380244e-05,
-      "loss": 3.1607,
+      "learning_rate": 4.9884011863929346e-05,
+      "loss": 3.1137,
       "step": 700
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.9857219856434565e-05,
-      "loss": 3.1491,
+      "learning_rate": 4.9867442130204974e-05,
+      "loss": 3.0948,
       "step": 800
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.9839372338488887e-05,
-      "loss": 3.141,
+      "learning_rate": 4.985087239648059e-05,
+      "loss": 3.0954,
       "step": 900
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.982152482054321e-05,
-      "loss": 3.1409,
+      "learning_rate": 4.983430266275621e-05,
+      "loss": 3.0936,
       "step": 1000
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.980367730259753e-05,
-      "loss": 3.1489,
+      "learning_rate": 4.981773292903183e-05,
+      "loss": 3.1028,
       "step": 1100
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.978582978465185e-05,
-      "loss": 3.1458,
+      "learning_rate": 4.980116319530745e-05,
+      "loss": 3.1106,
       "step": 1200
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.976798226670617e-05,
-      "loss": 3.1312,
+      "learning_rate": 4.9784593461583075e-05,
+      "loss": 3.098,
       "step": 1300
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.975013474876049e-05,
-      "loss": 3.1286,
+      "learning_rate": 4.9768023727858696e-05,
+      "loss": 3.091,
       "step": 1400
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.9732287230814814e-05,
-      "loss": 3.1277,
+      "epoch": 0.0,
+      "learning_rate": 4.975145399413432e-05,
+      "loss": 3.1045,
       "step": 1500
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9714439712869135e-05,
-      "loss": 3.1141,
+      "learning_rate": 4.973488426040994e-05,
+      "loss": 3.0708,
       "step": 1600
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.969659219492345e-05,
-      "loss": 3.1065,
+      "learning_rate": 4.971831452668556e-05,
+      "loss": 3.0879,
       "step": 1700
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.967874467697777e-05,
-      "loss": 3.1102,
+      "learning_rate": 4.970174479296118e-05,
+      "loss": 3.0888,
       "step": 1800
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.966089715903209e-05,
-      "loss": 3.1169,
+      "learning_rate": 4.96851750592368e-05,
+      "loss": 3.0854,
       "step": 1900
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.964304964108641e-05,
-      "loss": 3.115,
+      "learning_rate": 4.9668605325512425e-05,
+      "loss": 3.0843,
       "step": 2000
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9625202123140734e-05,
-      "loss": 3.1067,
+      "learning_rate": 4.9652035591788046e-05,
+      "loss": 3.0839,
       "step": 2100
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9607354605195056e-05,
-      "loss": 3.0965,
+      "learning_rate": 4.963546585806366e-05,
+      "loss": 3.0833,
       "step": 2200
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.958950708724938e-05,
-      "loss": 3.0902,
+      "learning_rate": 4.961889612433929e-05,
+      "loss": 3.0886,
       "step": 2300
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.95716595693037e-05,
-      "loss": 3.0859,
+      "learning_rate": 4.9602326390614904e-05,
+      "loss": 3.0796,
       "step": 2400
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.955381205135802e-05,
-      "loss": 3.0999,
+      "learning_rate": 4.9585756656890526e-05,
+      "loss": 3.0819,
       "step": 2500
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.953596453341234e-05,
-      "loss": 3.0828,
+      "learning_rate": 4.956918692316615e-05,
+      "loss": 3.0907,
       "step": 2600
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.951811701546666e-05,
-      "loss": 3.0801,
+      "learning_rate": 4.955261718944177e-05,
+      "loss": 3.0777,
       "step": 2700
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.950026949752098e-05,
-      "loss": 3.0769,
+      "learning_rate": 4.953604745571739e-05,
+      "loss": 3.0827,
       "step": 2800
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9482421979575304e-05,
-      "loss": 3.0826,
+      "learning_rate": 4.9519477721993005e-05,
+      "loss": 3.0713,
       "step": 2900
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9464574461629625e-05,
-      "loss": 3.0805,
+      "learning_rate": 4.950290798826863e-05,
+      "loss": 3.0912,
       "step": 3000
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.9446726943683947e-05,
-      "loss": 3.0767,
+      "learning_rate": 4.9486338254544254e-05,
+      "loss": 3.0765,
       "step": 3100
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.942887942573827e-05,
-      "loss": 3.0758,
+      "learning_rate": 4.946976852081987e-05,
+      "loss": 3.0899,
       "step": 3200
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.941103190779259e-05,
-      "loss": 3.0669,
+      "learning_rate": 4.94531987870955e-05,
+      "loss": 3.074,
       "step": 3300
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.939318438984691e-05,
-      "loss": 3.0613,
+      "learning_rate": 4.943662905337111e-05,
+      "loss": 3.0723,
       "step": 3400
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.937533687190123e-05,
-      "loss": 3.0743,
+      "learning_rate": 4.9420059319646733e-05,
+      "loss": 3.0737,
       "step": 3500
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.935748935395555e-05,
-      "loss": 3.0717,
+      "learning_rate": 4.9403489585922355e-05,
+      "loss": 3.0714,
       "step": 3600
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.933964183600987e-05,
-      "loss": 3.0649,
+      "learning_rate": 4.9386919852197976e-05,
+      "loss": 3.0751,
       "step": 3700
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.932179431806419e-05,
-      "loss": 3.0618,
+      "learning_rate": 4.93703501184736e-05,
+      "loss": 3.0701,
       "step": 3800
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.930394680011851e-05,
-      "loss": 3.0586,
+      "learning_rate": 4.935378038474922e-05,
+      "loss": 3.063,
       "step": 3900
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.928609928217283e-05,
-      "loss": 3.0575,
+      "learning_rate": 4.933721065102484e-05,
+      "loss": 3.071,
       "step": 4000
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.926825176422715e-05,
-      "loss": 3.0582,
+      "learning_rate": 4.932064091730046e-05,
+      "loss": 3.0686,
       "step": 4100
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.925040424628147e-05,
-      "loss": 3.0567,
+      "learning_rate": 4.9304071183576084e-05,
+      "loss": 3.0736,
       "step": 4200
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.9232556728335794e-05,
-      "loss": 3.0404,
+      "epoch": 0.01,
+      "learning_rate": 4.9287501449851705e-05,
+      "loss": 3.067,
       "step": 4300
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.9214709210390116e-05,
-      "loss": 3.0449,
+      "epoch": 0.01,
+      "learning_rate": 4.927093171612732e-05,
+      "loss": 3.072,
       "step": 4400
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 4.919686169244444e-05,
-      "loss": 3.0541,
+      "epoch": 0.01,
+      "learning_rate": 4.925436198240295e-05,
+      "loss": 3.0737,
       "step": 4500
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.917901417449876e-05,
-      "loss": 3.0451,
+      "learning_rate": 4.923779224867856e-05,
+      "loss": 3.0745,
       "step": 4600
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.916116665655308e-05,
-      "loss": 3.0483,
+      "learning_rate": 4.9221222514954184e-05,
+      "loss": 3.0572,
       "step": 4700
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.9143319138607394e-05,
-      "loss": 3.0485,
+      "learning_rate": 4.920465278122981e-05,
+      "loss": 3.0737,
       "step": 4800
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.9125471620661715e-05,
-      "loss": 3.0481,
+      "learning_rate": 4.918808304750543e-05,
+      "loss": 3.0674,
       "step": 4900
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.9107624102716036e-05,
-      "loss": 3.0357,
+      "learning_rate": 4.917151331378105e-05,
+      "loss": 3.0726,
       "step": 5000
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.908977658477036e-05,
-      "loss": 3.0463,
+      "learning_rate": 4.915494358005667e-05,
+      "loss": 3.0651,
       "step": 5100
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.907192906682468e-05,
-      "loss": 3.0369,
+      "learning_rate": 4.913837384633229e-05,
+      "loss": 3.0499,
       "step": 5200
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.9054081548879e-05,
-      "loss": 3.0465,
+      "learning_rate": 4.912180411260791e-05,
+      "loss": 3.0557,
       "step": 5300
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.903623403093332e-05,
-      "loss": 3.0229,
+      "learning_rate": 4.9105234378883535e-05,
+      "loss": 3.0773,
       "step": 5400
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.901838651298764e-05,
-      "loss": 3.0339,
+      "learning_rate": 4.9088664645159156e-05,
+      "loss": 3.0603,
       "step": 5500
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.900053899504196e-05,
-      "loss": 3.0449,
+      "learning_rate": 4.907209491143477e-05,
+      "loss": 3.0637,
       "step": 5600
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.8982691477096285e-05,
-      "loss": 3.0402,
+      "learning_rate": 4.90555251777104e-05,
+      "loss": 3.0585,
       "step": 5700
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.8964843959150606e-05,
-      "loss": 3.0287,
+      "learning_rate": 4.903895544398602e-05,
+      "loss": 3.0466,
       "step": 5800
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.894699644120492e-05,
-      "loss": 3.0269,
+      "learning_rate": 4.9022385710261635e-05,
+      "loss": 3.0489,
       "step": 5900
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.892914892325924e-05,
-      "loss": 3.0284,
+      "learning_rate": 4.900581597653726e-05,
+      "loss": 3.0555,
       "step": 6000
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.891130140531356e-05,
-      "loss": 3.0195,
+      "learning_rate": 4.898924624281288e-05,
+      "loss": 3.0683,
       "step": 6100
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.8893453887367884e-05,
-      "loss": 3.0309,
+      "learning_rate": 4.89726765090885e-05,
+      "loss": 3.0477,
       "step": 6200
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.8875606369422205e-05,
-      "loss": 3.0327,
+      "learning_rate": 4.895610677536413e-05,
+      "loss": 3.067,
       "step": 6300
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.8857758851476526e-05,
-      "loss": 3.0303,
+      "learning_rate": 4.893953704163974e-05,
+      "loss": 3.0541,
       "step": 6400
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.883991133353085e-05,
-      "loss": 3.0224,
+      "learning_rate": 4.8922967307915364e-05,
+      "loss": 3.0439,
       "step": 6500
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.882206381558517e-05,
-      "loss": 3.0304,
+      "learning_rate": 4.8906397574190985e-05,
+      "loss": 3.0552,
       "step": 6600
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.880421629763949e-05,
-      "loss": 3.0275,
+      "learning_rate": 4.888982784046661e-05,
+      "loss": 3.0491,
       "step": 6700
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.878636877969381e-05,
-      "loss": 3.0202,
+      "learning_rate": 4.887325810674223e-05,
+      "loss": 3.0408,
       "step": 6800
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.876852126174813e-05,
-      "loss": 3.0301,
+      "learning_rate": 4.885668837301785e-05,
+      "loss": 3.0552,
       "step": 6900
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.875067374380245e-05,
-      "loss": 3.0177,
+      "learning_rate": 4.884011863929347e-05,
+      "loss": 3.0526,
       "step": 7000
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.873282622585677e-05,
-      "loss": 3.0258,
+      "epoch": 0.02,
+      "learning_rate": 4.8823548905569086e-05,
+      "loss": 3.0611,
       "step": 7100
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.871497870791109e-05,
-      "loss": 3.0146,
+      "epoch": 0.02,
+      "learning_rate": 4.8806979171844714e-05,
+      "loss": 3.0663,
       "step": 7200
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.869713118996541e-05,
-      "loss": 3.0197,
+      "epoch": 0.02,
+      "learning_rate": 4.8790409438120336e-05,
+      "loss": 3.0376,
       "step": 7300
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.867928367201973e-05,
-      "loss": 3.0226,
+      "epoch": 0.02,
+      "learning_rate": 4.877383970439595e-05,
+      "loss": 3.0399,
       "step": 7400
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 4.866143615407405e-05,
-      "loss": 3.0154,
+      "epoch": 0.02,
+      "learning_rate": 4.875726997067158e-05,
+      "loss": 3.0532,
       "step": 7500
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8643588636128374e-05,
-      "loss": 3.0075,
+      "learning_rate": 4.874070023694719e-05,
+      "loss": 3.0499,
       "step": 7600
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.86257411181827e-05,
-      "loss": 3.0174,
+      "learning_rate": 4.8724130503222815e-05,
+      "loss": 3.0493,
       "step": 7700
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.860789360023702e-05,
-      "loss": 3.0117,
+      "learning_rate": 4.8707560769498436e-05,
+      "loss": 3.0654,
       "step": 7800
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.859004608229134e-05,
-      "loss": 3.0067,
+      "learning_rate": 4.869099103577406e-05,
+      "loss": 3.0431,
       "step": 7900
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.857219856434566e-05,
-      "loss": 3.0154,
+      "learning_rate": 4.867442130204968e-05,
+      "loss": 3.0464,
       "step": 8000
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.855435104639998e-05,
-      "loss": 3.016,
+      "learning_rate": 4.8657851568325294e-05,
+      "loss": 3.0636,
       "step": 8100
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.85365035284543e-05,
-      "loss": 3.0083,
+      "learning_rate": 4.864128183460092e-05,
+      "loss": 3.0482,
       "step": 8200
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.851865601050862e-05,
-      "loss": 3.0094,
+      "learning_rate": 4.8624712100876544e-05,
+      "loss": 3.044,
       "step": 8300
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8500808492562944e-05,
-      "loss": 3.0071,
+      "learning_rate": 4.860814236715216e-05,
+      "loss": 3.0512,
       "step": 8400
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8482960974617265e-05,
-      "loss": 3.0037,
+      "learning_rate": 4.8591572633427786e-05,
+      "loss": 3.0467,
       "step": 8500
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8465113456671586e-05,
-      "loss": 3.0166,
+      "learning_rate": 4.85750028997034e-05,
+      "loss": 3.0519,
       "step": 8600
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.844726593872591e-05,
-      "loss": 3.0083,
+      "learning_rate": 4.855843316597902e-05,
+      "loss": 3.0447,
       "step": 8700
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.842941842078023e-05,
-      "loss": 3.0019,
+      "learning_rate": 4.8541863432254644e-05,
+      "loss": 3.0328,
       "step": 8800
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.841157090283455e-05,
-      "loss": 3.0104,
+      "learning_rate": 4.8525293698530266e-05,
+      "loss": 3.044,
       "step": 8900
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8393723384888864e-05,
-      "loss": 2.9979,
+      "learning_rate": 4.850872396480589e-05,
+      "loss": 3.0545,
       "step": 9000
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8375875866943186e-05,
-      "loss": 3.0066,
+      "learning_rate": 4.849215423108151e-05,
+      "loss": 3.0425,
       "step": 9100
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.835802834899751e-05,
-      "loss": 3.0047,
+      "learning_rate": 4.847558449735713e-05,
+      "loss": 3.0544,
       "step": 9200
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.834018083105183e-05,
-      "loss": 3.0059,
+      "learning_rate": 4.845901476363275e-05,
+      "loss": 3.0411,
       "step": 9300
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.832233331310615e-05,
-      "loss": 2.9923,
+      "learning_rate": 4.844244502990837e-05,
+      "loss": 3.0471,
       "step": 9400
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.830448579516047e-05,
-      "loss": 2.9994,
+      "learning_rate": 4.8425875296183994e-05,
+      "loss": 3.0471,
       "step": 9500
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.828663827721479e-05,
-      "loss": 2.9912,
+      "learning_rate": 4.840930556245961e-05,
+      "loss": 3.0339,
       "step": 9600
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.826879075926911e-05,
-      "loss": 2.993,
+      "learning_rate": 4.839273582873524e-05,
+      "loss": 3.0515,
       "step": 9700
     },
     {
       "epoch": 0.03,
-      "learning_rate": 4.8250943241323434e-05,
-      "loss": 2.985,
+      "learning_rate": 4.837616609501085e-05,
+      "loss": 3.0327,
       "step": 9800
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.8233095723377755e-05,
-      "loss": 2.9926,
+      "epoch": 0.03,
+      "learning_rate": 4.8359596361286473e-05,
+      "loss": 3.0493,
       "step": 9900
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.8215248205432076e-05,
-      "loss": 2.9905,
+      "epoch": 0.03,
+      "learning_rate": 4.83430266275621e-05,
+      "loss": 3.0397,
       "step": 10000
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.819740068748639e-05,
-      "loss": 2.9819,
+      "epoch": 0.03,
+      "learning_rate": 4.8326456893837716e-05,
+      "loss": 3.0343,
       "step": 10100
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.817955316954071e-05,
-      "loss": 2.9921,
+      "epoch": 0.03,
+      "learning_rate": 4.830988716011334e-05,
+      "loss": 3.0395,
       "step": 10200
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.816170565159503e-05,
-      "loss": 2.9812,
+      "epoch": 0.03,
+      "learning_rate": 4.829331742638896e-05,
+      "loss": 3.0369,
       "step": 10300
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.8143858133649355e-05,
-      "loss": 2.9954,
+      "epoch": 0.03,
+      "learning_rate": 4.827674769266458e-05,
+      "loss": 3.0426,
       "step": 10400
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.8126010615703676e-05,
-      "loss": 2.9938,
+      "epoch": 0.03,
+      "learning_rate": 4.82601779589402e-05,
+      "loss": 3.0444,
       "step": 10500
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.8108163097758e-05,
-      "loss": 2.983,
+      "learning_rate": 4.8243608225215824e-05,
+      "loss": 3.0418,
       "step": 10600
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.809031557981232e-05,
-      "loss": 2.9933,
+      "learning_rate": 4.8227038491491445e-05,
+      "loss": 3.0404,
       "step": 10700
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.807246806186664e-05,
-      "loss": 2.9954,
+      "learning_rate": 4.821046875776707e-05,
+      "loss": 3.0291,
       "step": 10800
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.805462054392096e-05,
-      "loss": 2.9773,
+      "learning_rate": 4.819389902404269e-05,
+      "loss": 3.0299,
       "step": 10900
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.803677302597528e-05,
-      "loss": 2.9905,
+      "learning_rate": 4.817732929031831e-05,
+      "loss": 3.0373,
       "step": 11000
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.80189255080296e-05,
-      "loss": 2.9884,
+      "learning_rate": 4.8160759556593924e-05,
+      "loss": 3.0301,
       "step": 11100
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.800107799008392e-05,
-      "loss": 2.9752,
+      "learning_rate": 4.814418982286955e-05,
+      "loss": 3.0319,
       "step": 11200
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.798323047213824e-05,
-      "loss": 2.9862,
+      "learning_rate": 4.812762008914517e-05,
+      "loss": 3.0302,
       "step": 11300
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.796538295419256e-05,
-      "loss": 2.983,
+      "learning_rate": 4.811105035542079e-05,
+      "loss": 3.039,
       "step": 11400
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.794753543624688e-05,
-      "loss": 2.9796,
+      "learning_rate": 4.809448062169642e-05,
+      "loss": 3.0367,
       "step": 11500
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.79296879183012e-05,
-      "loss": 2.9725,
+      "learning_rate": 4.807791088797203e-05,
+      "loss": 3.0371,
       "step": 11600
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.7911840400355524e-05,
-      "loss": 2.9862,
+      "learning_rate": 4.806134115424765e-05,
+      "loss": 3.0406,
       "step": 11700
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.7893992882409845e-05,
-      "loss": 2.9857,
+      "learning_rate": 4.8044771420523275e-05,
+      "loss": 3.0481,
       "step": 11800
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.7876145364464166e-05,
-      "loss": 2.9709,
+      "learning_rate": 4.8028201686798896e-05,
+      "loss": 3.0449,
       "step": 11900
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.785829784651849e-05,
-      "loss": 2.9848,
+      "learning_rate": 4.801163195307452e-05,
+      "loss": 3.0242,
       "step": 12000
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.784045032857281e-05,
-      "loss": 2.9783,
+      "learning_rate": 4.799506221935014e-05,
+      "loss": 3.0403,
       "step": 12100
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.782260281062712e-05,
-      "loss": 2.9718,
+      "learning_rate": 4.797849248562576e-05,
+      "loss": 3.0297,
       "step": 12200
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.7804755292681444e-05,
-      "loss": 2.9787,
+      "learning_rate": 4.7961922751901375e-05,
+      "loss": 3.0259,
       "step": 12300
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.7786907774735765e-05,
-      "loss": 2.9676,
+      "learning_rate": 4.7945353018177e-05,
+      "loss": 3.048,
       "step": 12400
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.7769060256790086e-05,
-      "loss": 2.9666,
+      "learning_rate": 4.7928783284452625e-05,
+      "loss": 3.0273,
       "step": 12500
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.775121273884441e-05,
-      "loss": 2.9778,
+      "learning_rate": 4.791221355072824e-05,
+      "loss": 3.0195,
       "step": 12600
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.773336522089873e-05,
-      "loss": 2.9741,
+      "epoch": 0.04,
+      "learning_rate": 4.789564381700387e-05,
+      "loss": 3.0402,
       "step": 12700
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.771551770295306e-05,
-      "loss": 2.9708,
+      "epoch": 0.04,
+      "learning_rate": 4.787907408327948e-05,
+      "loss": 3.0298,
       "step": 12800
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.769767018500738e-05,
-      "loss": 2.9686,
+      "epoch": 0.04,
+      "learning_rate": 4.7862504349555104e-05,
+      "loss": 3.0353,
       "step": 12900
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.76798226670617e-05,
-      "loss": 2.9707,
+      "epoch": 0.04,
+      "learning_rate": 4.7845934615830725e-05,
+      "loss": 3.016,
       "step": 13000
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.7661975149116014e-05,
-      "loss": 2.9682,
+      "epoch": 0.04,
+      "learning_rate": 4.782936488210635e-05,
+      "loss": 3.0316,
       "step": 13100
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.7644127631170335e-05,
-      "loss": 2.966,
+      "epoch": 0.04,
+      "learning_rate": 4.781279514838197e-05,
+      "loss": 3.0273,
       "step": 13200
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.7626280113224656e-05,
-      "loss": 2.9716,
+      "epoch": 0.04,
+      "learning_rate": 4.779622541465758e-05,
+      "loss": 3.0323,
       "step": 13300
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.760843259527898e-05,
-      "loss": 2.9772,
+      "epoch": 0.04,
+      "learning_rate": 4.777965568093321e-05,
+      "loss": 3.0267,
       "step": 13400
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 4.75905850773333e-05,
-      "loss": 2.9659,
+      "epoch": 0.04,
+      "learning_rate": 4.776308594720883e-05,
+      "loss": 3.0375,
       "step": 13500
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.757273755938762e-05,
-      "loss": 2.9753,
+      "learning_rate": 4.774651621348445e-05,
+      "loss": 3.0289,
       "step": 13600
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.755489004144194e-05,
-      "loss": 2.9617,
+      "learning_rate": 4.7729946479760076e-05,
+      "loss": 3.0271,
       "step": 13700
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.753704252349626e-05,
-      "loss": 2.9614,
+      "learning_rate": 4.771337674603569e-05,
+      "loss": 3.0204,
       "step": 13800
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7519195005550584e-05,
-      "loss": 2.9713,
+      "learning_rate": 4.769680701231131e-05,
+      "loss": 3.0219,
       "step": 13900
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7501347487604905e-05,
-      "loss": 2.9694,
+      "learning_rate": 4.768023727858694e-05,
+      "loss": 3.0396,
       "step": 14000
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7483499969659226e-05,
-      "loss": 2.9663,
+      "learning_rate": 4.7663667544862555e-05,
+      "loss": 3.023,
       "step": 14100
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.746565245171354e-05,
-      "loss": 2.9633,
+      "learning_rate": 4.7647097811138176e-05,
+      "loss": 3.0274,
       "step": 14200
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.744780493376786e-05,
-      "loss": 2.9668,
+      "learning_rate": 4.76305280774138e-05,
+      "loss": 3.0362,
       "step": 14300
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.742995741582218e-05,
-      "loss": 2.956,
+      "learning_rate": 4.761395834368942e-05,
+      "loss": 3.0186,
       "step": 14400
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7412109897876504e-05,
-      "loss": 2.9591,
+      "learning_rate": 4.759738860996504e-05,
+      "loss": 3.0279,
       "step": 14500
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7394262379930825e-05,
-      "loss": 2.9598,
+      "learning_rate": 4.758081887624066e-05,
+      "loss": 3.0242,
       "step": 14600
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7376414861985146e-05,
-      "loss": 2.95,
+      "learning_rate": 4.7564249142516284e-05,
+      "loss": 3.0171,
       "step": 14700
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.735856734403947e-05,
-      "loss": 2.9574,
+      "learning_rate": 4.75476794087919e-05,
+      "loss": 3.0135,
       "step": 14800
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.734071982609379e-05,
-      "loss": 2.9655,
+      "learning_rate": 4.7531109675067526e-05,
+      "loss": 3.0179,
       "step": 14900
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.732287230814811e-05,
-      "loss": 2.9585,
+      "learning_rate": 4.751453994134315e-05,
+      "loss": 3.0292,
       "step": 15000
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.730502479020243e-05,
-      "loss": 2.9571,
+      "learning_rate": 4.749797020761876e-05,
+      "loss": 3.036,
       "step": 15100
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.728717727225675e-05,
-      "loss": 2.9675,
+      "learning_rate": 4.748140047389439e-05,
+      "loss": 3.0231,
       "step": 15200
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.726932975431107e-05,
-      "loss": 2.9577,
+      "learning_rate": 4.7464830740170006e-05,
+      "loss": 3.0235,
       "step": 15300
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.725148223636539e-05,
-      "loss": 2.9613,
+      "learning_rate": 4.744826100644563e-05,
+      "loss": 3.0347,
       "step": 15400
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.723363471841971e-05,
-      "loss": 2.9628,
+      "epoch": 0.05,
+      "learning_rate": 4.743169127272125e-05,
+      "loss": 3.0356,
       "step": 15500
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.721578720047403e-05,
-      "loss": 2.9584,
+      "epoch": 0.05,
+      "learning_rate": 4.741512153899687e-05,
+      "loss": 3.0229,
       "step": 15600
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.719793968252835e-05,
-      "loss": 2.9501,
+      "epoch": 0.05,
+      "learning_rate": 4.739855180527249e-05,
+      "loss": 3.0371,
       "step": 15700
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.718009216458267e-05,
-      "loss": 2.9566,
+      "epoch": 0.05,
+      "learning_rate": 4.738198207154811e-05,
+      "loss": 3.0207,
       "step": 15800
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.7162244646636994e-05,
-      "loss": 2.9547,
+      "epoch": 0.05,
+      "learning_rate": 4.7365412337823734e-05,
+      "loss": 3.028,
       "step": 15900
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.7144397128691315e-05,
-      "loss": 2.9563,
+      "epoch": 0.05,
+      "learning_rate": 4.7348842604099356e-05,
+      "loss": 3.0201,
       "step": 16000
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.712654961074564e-05,
-      "loss": 2.9514,
+      "epoch": 0.05,
+      "learning_rate": 4.733227287037498e-05,
+      "loss": 3.0075,
       "step": 16100
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.710870209279996e-05,
-      "loss": 2.953,
+      "epoch": 0.05,
+      "learning_rate": 4.73157031366506e-05,
+      "loss": 3.0233,
       "step": 16200
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.709085457485428e-05,
-      "loss": 2.9614,
+      "epoch": 0.05,
+      "learning_rate": 4.7299133402926213e-05,
+      "loss": 3.0093,
       "step": 16300
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.7073007056908594e-05,
-      "loss": 2.9524,
+      "epoch": 0.05,
+      "learning_rate": 4.728256366920184e-05,
+      "loss": 3.026,
       "step": 16400
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 4.7055159538962915e-05,
-      "loss": 2.9636,
+      "epoch": 0.05,
+      "learning_rate": 4.7265993935477456e-05,
+      "loss": 3.0331,
       "step": 16500
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.7037312021017236e-05,
-      "loss": 2.9469,
+      "learning_rate": 4.724942420175308e-05,
+      "loss": 3.0273,
       "step": 16600
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.701946450307156e-05,
-      "loss": 2.9505,
+      "learning_rate": 4.7232854468028706e-05,
+      "loss": 3.0141,
       "step": 16700
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.700161698512588e-05,
-      "loss": 2.9413,
+      "learning_rate": 4.721628473430432e-05,
+      "loss": 3.0135,
       "step": 16800
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.69837694671802e-05,
-      "loss": 2.9437,
+      "learning_rate": 4.719971500057994e-05,
+      "loss": 3.0046,
       "step": 16900
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.696592194923452e-05,
-      "loss": 2.9568,
+      "learning_rate": 4.7183145266855564e-05,
+      "loss": 3.0226,
       "step": 17000
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.694807443128884e-05,
-      "loss": 2.9489,
+      "learning_rate": 4.7166575533131185e-05,
+      "loss": 3.0071,
       "step": 17100
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.693022691334316e-05,
-      "loss": 2.9461,
+      "learning_rate": 4.715000579940681e-05,
+      "loss": 3.0271,
       "step": 17200
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.6912379395397484e-05,
-      "loss": 2.9527,
+      "learning_rate": 4.713343606568243e-05,
+      "loss": 3.0136,
       "step": 17300
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.6894531877451806e-05,
-      "loss": 2.9453,
+      "learning_rate": 4.711686633195805e-05,
+      "loss": 3.0247,
       "step": 17400
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.687668435950612e-05,
-      "loss": 2.9504,
+      "learning_rate": 4.7100296598233664e-05,
+      "loss": 3.0245,
       "step": 17500
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.685883684156044e-05,
-      "loss": 2.9493,
+      "learning_rate": 4.708372686450929e-05,
+      "loss": 2.9972,
       "step": 17600
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.684098932361476e-05,
-      "loss": 2.9494,
+      "learning_rate": 4.7067157130784914e-05,
+      "loss": 3.0169,
       "step": 17700
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.6823141805669084e-05,
-      "loss": 2.9507,
+      "learning_rate": 4.705058739706053e-05,
+      "loss": 3.0187,
       "step": 17800
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.680529428772341e-05,
-      "loss": 2.9527,
+      "learning_rate": 4.703401766333616e-05,
+      "loss": 3.0095,
       "step": 17900
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.678744676977773e-05,
-      "loss": 2.9302,
+      "learning_rate": 4.701744792961177e-05,
+      "loss": 3.0072,
       "step": 18000
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.6769599251832054e-05,
-      "loss": 2.9517,
+      "learning_rate": 4.700087819588739e-05,
+      "loss": 3.0188,
       "step": 18100
     },
     {
       "epoch": 0.06,
-      "learning_rate": 4.6751751733886375e-05,
-      "loss": 2.927,
+      "learning_rate": 4.6984308462163015e-05,
+      "loss": 3.0217,
       "step": 18200
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.67339042159407e-05,
-      "loss": 2.9401,
+      "epoch": 0.06,
+      "learning_rate": 4.6967738728438636e-05,
+      "loss": 3.0296,
       "step": 18300
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.671605669799501e-05,
-      "loss": 2.9564,
+      "epoch": 0.06,
+      "learning_rate": 4.695116899471426e-05,
+      "loss": 3.0292,
       "step": 18400
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.669820918004933e-05,
-      "loss": 2.943,
+      "epoch": 0.06,
+      "learning_rate": 4.693459926098987e-05,
+      "loss": 3.0092,
       "step": 18500
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.6680361662103654e-05,
-      "loss": 2.9414,
+      "epoch": 0.06,
+      "learning_rate": 4.69180295272655e-05,
+      "loss": 3.0257,
       "step": 18600
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.6662514144157975e-05,
-      "loss": 2.9446,
+      "epoch": 0.06,
+      "learning_rate": 4.690145979354112e-05,
+      "loss": 3.0249,
       "step": 18700
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.6644666626212296e-05,
-      "loss": 2.9462,
+      "epoch": 0.06,
+      "learning_rate": 4.6884890059816737e-05,
+      "loss": 3.0094,
       "step": 18800
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.662681910826662e-05,
-      "loss": 2.9353,
+      "epoch": 0.06,
+      "learning_rate": 4.6868320326092365e-05,
+      "loss": 2.9962,
       "step": 18900
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.660897159032094e-05,
-      "loss": 2.9467,
+      "epoch": 0.06,
+      "learning_rate": 4.685175059236798e-05,
+      "loss": 3.0184,
       "step": 19000
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.659112407237526e-05,
-      "loss": 2.9301,
+      "epoch": 0.06,
+      "learning_rate": 4.68351808586436e-05,
+      "loss": 3.005,
       "step": 19100
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.657327655442958e-05,
-      "loss": 2.9428,
+      "epoch": 0.06,
+      "learning_rate": 4.681861112491923e-05,
+      "loss": 3.0167,
       "step": 19200
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.65554290364839e-05,
-      "loss": 2.9403,
+      "epoch": 0.06,
+      "learning_rate": 4.6802041391194844e-05,
+      "loss": 3.0212,
       "step": 19300
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.653758151853822e-05,
-      "loss": 2.9301,
+      "epoch": 0.06,
+      "learning_rate": 4.6785471657470465e-05,
+      "loss": 2.9941,
       "step": 19400
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.651973400059254e-05,
-      "loss": 2.9338,
+      "epoch": 0.06,
+      "learning_rate": 4.676890192374609e-05,
+      "loss": 3.0055,
       "step": 19500
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 4.650188648264686e-05,
-      "loss": 2.9398,
+      "epoch": 0.06,
+      "learning_rate": 4.675233219002171e-05,
+      "loss": 3.0132,
       "step": 19600
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.648403896470118e-05,
-      "loss": 2.9341,
+      "learning_rate": 4.673576245629733e-05,
+      "loss": 3.0185,
       "step": 19700
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.64661914467555e-05,
-      "loss": 2.9378,
+      "learning_rate": 4.671919272257295e-05,
+      "loss": 3.0065,
       "step": 19800
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.644834392880982e-05,
-      "loss": 2.9242,
+      "learning_rate": 4.670262298884857e-05,
+      "loss": 2.9932,
       "step": 19900
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.6430496410864144e-05,
-      "loss": 2.9369,
+      "learning_rate": 4.668605325512419e-05,
+      "loss": 3.0123,
       "step": 20000
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.6412648892918465e-05,
-      "loss": 2.9442,
+      "learning_rate": 4.6669483521399816e-05,
+      "loss": 3.0023,
       "step": 20100
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.6394801374972786e-05,
-      "loss": 2.934,
+      "learning_rate": 4.665291378767544e-05,
+      "loss": 3.0135,
       "step": 20200
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.637695385702711e-05,
-      "loss": 2.9385,
+      "learning_rate": 4.663634405395105e-05,
+      "loss": 3.0097,
       "step": 20300
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.635910633908143e-05,
-      "loss": 2.9304,
+      "learning_rate": 4.661977432022668e-05,
+      "loss": 3.0129,
       "step": 20400
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.634125882113575e-05,
-      "loss": 2.9351,
+      "learning_rate": 4.6603204586502295e-05,
+      "loss": 3.0222,
       "step": 20500
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.6323411303190064e-05,
-      "loss": 2.9297,
+      "learning_rate": 4.6586634852777916e-05,
+      "loss": 3.0131,
       "step": 20600
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.6305563785244385e-05,
-      "loss": 2.9271,
+      "learning_rate": 4.657006511905354e-05,
+      "loss": 3.0084,
       "step": 20700
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.628771626729871e-05,
-      "loss": 2.9317,
+      "learning_rate": 4.655349538532916e-05,
+      "loss": 3.0145,
       "step": 20800
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.626986874935303e-05,
-      "loss": 2.9295,
+      "learning_rate": 4.653692565160478e-05,
+      "loss": 3.001,
       "step": 20900
     },
     {
       "epoch": 0.07,
-      "learning_rate": 4.625202123140735e-05,
-      "loss": 2.9402,
+      "learning_rate": 4.65203559178804e-05,
+      "loss": 3.0132,
       "step": 21000
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.623417371346167e-05,
-      "loss": 2.9259,
+      "epoch": 0.07,
+      "learning_rate": 4.6503786184156024e-05,
+      "loss": 3.0006,
       "step": 21100
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.621632619551599e-05,
-      "loss": 2.933,
+      "epoch": 0.07,
+      "learning_rate": 4.6487216450431645e-05,
+      "loss": 2.9896,
       "step": 21200
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.619847867757031e-05,
-      "loss": 2.9216,
+      "epoch": 0.07,
+      "learning_rate": 4.6470646716707266e-05,
+      "loss": 3.015,
       "step": 21300
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6180631159624634e-05,
-      "loss": 2.933,
+      "epoch": 0.07,
+      "learning_rate": 4.645407698298289e-05,
+      "loss": 3.0095,
       "step": 21400
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6162783641678955e-05,
-      "loss": 2.9376,
+      "epoch": 0.07,
+      "learning_rate": 4.64375072492585e-05,
+      "loss": 3.0047,
       "step": 21500
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6144936123733276e-05,
-      "loss": 2.9255,
+      "epoch": 0.07,
+      "learning_rate": 4.642093751553413e-05,
+      "loss": 3.0002,
       "step": 21600
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.612708860578759e-05,
-      "loss": 2.9342,
+      "epoch": 0.07,
+      "learning_rate": 4.6404367781809746e-05,
+      "loss": 3.0043,
       "step": 21700
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.610924108784191e-05,
-      "loss": 2.9278,
+      "epoch": 0.07,
+      "learning_rate": 4.638779804808537e-05,
+      "loss": 3.0,
       "step": 21800
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.609139356989623e-05,
-      "loss": 2.9181,
+      "epoch": 0.07,
+      "learning_rate": 4.6371228314360995e-05,
+      "loss": 3.0076,
       "step": 21900
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6073546051950554e-05,
-      "loss": 2.927,
+      "epoch": 0.07,
+      "learning_rate": 4.635465858063661e-05,
+      "loss": 3.0087,
       "step": 22000
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.6055698534004876e-05,
-      "loss": 2.9297,
+      "epoch": 0.07,
+      "learning_rate": 4.633808884691223e-05,
+      "loss": 3.0113,
       "step": 22100
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.60378510160592e-05,
-      "loss": 2.9301,
+      "epoch": 0.07,
+      "learning_rate": 4.632151911318785e-05,
+      "loss": 3.0093,
       "step": 22200
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.602000349811352e-05,
-      "loss": 2.9257,
+      "epoch": 0.07,
+      "learning_rate": 4.6304949379463474e-05,
+      "loss": 3.0017,
       "step": 22300
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.600215598016784e-05,
-      "loss": 2.9263,
+      "epoch": 0.07,
+      "learning_rate": 4.6288379645739096e-05,
+      "loss": 2.9954,
       "step": 22400
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.598430846222216e-05,
-      "loss": 2.9152,
+      "epoch": 0.07,
+      "learning_rate": 4.627180991201472e-05,
+      "loss": 3.0096,
       "step": 22500
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.596646094427648e-05,
-      "loss": 2.9272,
+      "epoch": 0.07,
+      "learning_rate": 4.625524017829034e-05,
+      "loss": 2.9921,
       "step": 22600
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.59486134263308e-05,
-      "loss": 2.9196,
+      "learning_rate": 4.623867044456596e-05,
+      "loss": 2.9997,
       "step": 22700
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.593076590838512e-05,
-      "loss": 2.9285,
+      "learning_rate": 4.622210071084158e-05,
+      "loss": 2.9966,
       "step": 22800
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.591291839043944e-05,
-      "loss": 2.9176,
+      "learning_rate": 4.62055309771172e-05,
+      "loss": 2.9913,
       "step": 22900
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.589507087249377e-05,
-      "loss": 2.9267,
+      "learning_rate": 4.618896124339282e-05,
+      "loss": 3.0022,
       "step": 23000
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.587722335454809e-05,
-      "loss": 2.9226,
+      "learning_rate": 4.6172391509668446e-05,
+      "loss": 3.0091,
       "step": 23100
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.585937583660241e-05,
-      "loss": 2.9278,
+      "learning_rate": 4.615582177594406e-05,
+      "loss": 3.0005,
       "step": 23200
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.584152831865673e-05,
-      "loss": 2.9186,
+      "learning_rate": 4.613925204221968e-05,
+      "loss": 3.001,
       "step": 23300
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.582368080071105e-05,
-      "loss": 2.9262,
+      "learning_rate": 4.6122682308495304e-05,
+      "loss": 3.005,
       "step": 23400
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.580583328276537e-05,
-      "loss": 2.9204,
+      "learning_rate": 4.6106112574770925e-05,
+      "loss": 2.9926,
       "step": 23500
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.5787985764819694e-05,
-      "loss": 2.9287,
+      "learning_rate": 4.608954284104655e-05,
+      "loss": 2.9994,
       "step": 23600
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.577013824687401e-05,
-      "loss": 2.9274,
+      "learning_rate": 4.607297310732217e-05,
+      "loss": 2.9934,
       "step": 23700
     },
     {
       "epoch": 0.08,
-      "learning_rate": 4.575229072892833e-05,
-      "loss": 2.904,
+      "learning_rate": 4.605640337359779e-05,
+      "loss": 2.9978,
       "step": 23800
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.573444321098265e-05,
-      "loss": 2.9017,
+      "epoch": 0.08,
+      "learning_rate": 4.603983363987341e-05,
+      "loss": 3.0013,
       "step": 23900
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.571659569303697e-05,
-      "loss": 2.9142,
+      "epoch": 0.08,
+      "learning_rate": 4.6023263906149026e-05,
+      "loss": 2.9865,
       "step": 24000
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.569874817509129e-05,
-      "loss": 2.9195,
+      "epoch": 0.08,
+      "learning_rate": 4.6006694172424654e-05,
+      "loss": 2.9939,
       "step": 24100
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5680900657145614e-05,
-      "loss": 2.9171,
+      "epoch": 0.08,
+      "learning_rate": 4.599012443870027e-05,
+      "loss": 3.0167,
       "step": 24200
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5663053139199936e-05,
-      "loss": 2.921,
+      "epoch": 0.08,
+      "learning_rate": 4.597355470497589e-05,
+      "loss": 3.0104,
       "step": 24300
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.564520562125426e-05,
-      "loss": 2.9243,
+      "epoch": 0.08,
+      "learning_rate": 4.595698497125152e-05,
+      "loss": 2.9989,
       "step": 24400
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.562735810330858e-05,
-      "loss": 2.9245,
+      "epoch": 0.08,
+      "learning_rate": 4.594041523752713e-05,
+      "loss": 2.9975,
       "step": 24500
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.56095105853629e-05,
-      "loss": 2.9135,
+      "epoch": 0.08,
+      "learning_rate": 4.5923845503802755e-05,
+      "loss": 2.9963,
       "step": 24600
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.559166306741722e-05,
-      "loss": 2.9224,
+      "epoch": 0.08,
+      "learning_rate": 4.5907275770078376e-05,
+      "loss": 2.997,
       "step": 24700
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5573815549471535e-05,
-      "loss": 2.9328,
+      "epoch": 0.08,
+      "learning_rate": 4.5890706036354e-05,
+      "loss": 3.0045,
       "step": 24800
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5555968031525856e-05,
-      "loss": 2.9171,
+      "epoch": 0.08,
+      "learning_rate": 4.587413630262962e-05,
+      "loss": 2.99,
       "step": 24900
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.553812051358018e-05,
-      "loss": 2.9026,
+      "epoch": 0.08,
+      "learning_rate": 4.585756656890524e-05,
+      "loss": 2.9822,
       "step": 25000
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.55202729956345e-05,
-      "loss": 2.9174,
+      "epoch": 0.08,
+      "learning_rate": 4.584099683518086e-05,
+      "loss": 3.0053,
       "step": 25100
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.550242547768882e-05,
-      "loss": 2.915,
+      "epoch": 0.08,
+      "learning_rate": 4.5824427101456477e-05,
+      "loss": 3.0055,
       "step": 25200
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.548457795974314e-05,
-      "loss": 2.9166,
+      "epoch": 0.08,
+      "learning_rate": 4.5807857367732105e-05,
+      "loss": 3.0025,
       "step": 25300
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.546673044179746e-05,
-      "loss": 2.92,
+      "epoch": 0.08,
+      "learning_rate": 4.5791287634007726e-05,
+      "loss": 3.0042,
       "step": 25400
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5448882923851783e-05,
-      "loss": 2.9208,
+      "epoch": 0.08,
+      "learning_rate": 4.577471790028334e-05,
+      "loss": 3.0041,
       "step": 25500
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5431035405906105e-05,
-      "loss": 2.9125,
+      "epoch": 0.08,
+      "learning_rate": 4.575814816655897e-05,
+      "loss": 3.001,
       "step": 25600
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.5413187887960426e-05,
-      "loss": 2.9141,
+      "learning_rate": 4.5741578432834584e-05,
+      "loss": 2.9987,
       "step": 25700
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.539534037001475e-05,
-      "loss": 2.9063,
+      "learning_rate": 4.5725008699110205e-05,
+      "loss": 2.9917,
       "step": 25800
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.537749285206906e-05,
-      "loss": 2.9108,
+      "learning_rate": 4.5708438965385834e-05,
+      "loss": 3.0052,
       "step": 25900
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.535964533412338e-05,
-      "loss": 2.8963,
+      "learning_rate": 4.569186923166145e-05,
+      "loss": 2.9966,
       "step": 26000
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.5341797816177704e-05,
-      "loss": 2.9079,
+      "learning_rate": 4.567529949793707e-05,
+      "loss": 2.9915,
       "step": 26100
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.5323950298232025e-05,
-      "loss": 2.9122,
+      "learning_rate": 4.565872976421269e-05,
+      "loss": 3.0024,
       "step": 26200
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.5306102780286346e-05,
-      "loss": 2.9094,
+      "learning_rate": 4.564216003048831e-05,
+      "loss": 2.9882,
       "step": 26300
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.528825526234067e-05,
-      "loss": 2.9195,
+      "learning_rate": 4.5625590296763934e-05,
+      "loss": 3.0081,
       "step": 26400
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.527040774439499e-05,
-      "loss": 2.9079,
+      "learning_rate": 4.5609020563039556e-05,
+      "loss": 2.9974,
       "step": 26500
     },
     {
       "epoch": 0.09,
-      "learning_rate": 4.525256022644931e-05,
-      "loss": 2.9167,
+      "learning_rate": 4.559245082931518e-05,
+      "loss": 2.9949,
       "step": 26600
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.523471270850363e-05,
-      "loss": 2.9091,
+      "epoch": 0.09,
+      "learning_rate": 4.557588109559079e-05,
+      "loss": 2.997,
       "step": 26700
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.521686519055795e-05,
-      "loss": 2.9022,
+      "epoch": 0.09,
+      "learning_rate": 4.555931136186642e-05,
+      "loss": 2.9953,
       "step": 26800
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.5199017672612274e-05,
-      "loss": 2.9099,
+      "epoch": 0.09,
+      "learning_rate": 4.554274162814204e-05,
+      "loss": 2.993,
       "step": 26900
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.518117015466659e-05,
-      "loss": 2.9264,
+      "epoch": 0.09,
+      "learning_rate": 4.5526171894417656e-05,
+      "loss": 2.9972,
       "step": 27000
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.516332263672091e-05,
-      "loss": 2.911,
+      "epoch": 0.09,
+      "learning_rate": 4.5509602160693284e-05,
+      "loss": 2.9947,
       "step": 27100
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.514547511877523e-05,
-      "loss": 2.9119,
+      "epoch": 0.09,
+      "learning_rate": 4.54930324269689e-05,
+      "loss": 3.0025,
       "step": 27200
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.512762760082955e-05,
-      "loss": 2.9108,
+      "epoch": 0.09,
+      "learning_rate": 4.547646269324452e-05,
+      "loss": 3.0024,
       "step": 27300
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.510978008288387e-05,
-      "loss": 2.9135,
+      "epoch": 0.09,
+      "learning_rate": 4.545989295952014e-05,
+      "loss": 2.993,
       "step": 27400
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.5091932564938194e-05,
-      "loss": 2.9107,
+      "epoch": 0.09,
+      "learning_rate": 4.5443323225795764e-05,
+      "loss": 3.001,
       "step": 27500
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.5074085046992515e-05,
-      "loss": 2.8936,
+      "epoch": 0.09,
+      "learning_rate": 4.5426753492071385e-05,
+      "loss": 3.002,
       "step": 27600
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.505623752904684e-05,
-      "loss": 2.9127,
+      "epoch": 0.09,
+      "learning_rate": 4.5410183758347006e-05,
+      "loss": 2.9877,
       "step": 27700
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.503839001110116e-05,
-      "loss": 2.9002,
+      "epoch": 0.09,
+      "learning_rate": 4.539361402462263e-05,
+      "loss": 2.9944,
       "step": 27800
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.502054249315548e-05,
-      "loss": 2.9018,
+      "epoch": 0.09,
+      "learning_rate": 4.537704429089825e-05,
+      "loss": 3.0094,
       "step": 27900
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.50026949752098e-05,
-      "loss": 2.8991,
+      "epoch": 0.09,
+      "learning_rate": 4.536047455717387e-05,
+      "loss": 2.9904,
       "step": 28000
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.498484745726412e-05,
-      "loss": 2.9029,
+      "epoch": 0.09,
+      "learning_rate": 4.534390482344949e-05,
+      "loss": 2.9992,
       "step": 28100
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.496699993931844e-05,
-      "loss": 2.9135,
+      "epoch": 0.09,
+      "learning_rate": 4.532733508972511e-05,
+      "loss": 2.9974,
       "step": 28200
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.4949152421372764e-05,
-      "loss": 2.9018,
+      "epoch": 0.09,
+      "learning_rate": 4.5310765356000735e-05,
+      "loss": 2.9998,
       "step": 28300
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.4931304903427085e-05,
-      "loss": 2.9132,
+      "epoch": 0.09,
+      "learning_rate": 4.529419562227635e-05,
+      "loss": 2.9959,
       "step": 28400
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.4913457385481406e-05,
-      "loss": 2.915,
+      "epoch": 0.09,
+      "learning_rate": 4.527762588855197e-05,
+      "loss": 2.9885,
       "step": 28500
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.489560986753573e-05,
-      "loss": 2.8909,
+      "epoch": 0.09,
+      "learning_rate": 4.526105615482759e-05,
+      "loss": 2.9884,
       "step": 28600
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.487776234959005e-05,
-      "loss": 2.9024,
+      "learning_rate": 4.5244486421103214e-05,
+      "loss": 2.9907,
       "step": 28700
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.485991483164437e-05,
-      "loss": 2.8926,
+      "learning_rate": 4.5227916687378836e-05,
+      "loss": 2.9801,
       "step": 28800
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.484206731369869e-05,
-      "loss": 2.8982,
+      "learning_rate": 4.521134695365446e-05,
+      "loss": 2.9874,
       "step": 28900
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.4824219795753006e-05,
-      "loss": 2.9044,
+      "learning_rate": 4.519477721993008e-05,
+      "loss": 2.9943,
       "step": 29000
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.480637227780733e-05,
-      "loss": 2.9007,
+      "learning_rate": 4.51782074862057e-05,
+      "loss": 2.9778,
       "step": 29100
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.478852475986165e-05,
-      "loss": 2.8998,
+      "learning_rate": 4.5161637752481315e-05,
+      "loss": 3.0045,
       "step": 29200
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.477067724191597e-05,
-      "loss": 2.9021,
+      "learning_rate": 4.514506801875694e-05,
+      "loss": 2.986,
       "step": 29300
     },
     {
       "epoch": 0.1,
-      "learning_rate": 4.475282972397029e-05,
-      "loss": 2.9067,
+      "learning_rate": 4.512849828503256e-05,
+      "loss": 2.9972,
       "step": 29400
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.473498220602461e-05,
-      "loss": 2.9029,
+      "epoch": 0.1,
+      "learning_rate": 4.511192855130818e-05,
+      "loss": 2.9938,
       "step": 29500
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.471713468807893e-05,
-      "loss": 2.8996,
+      "epoch": 0.1,
+      "learning_rate": 4.509535881758381e-05,
+      "loss": 2.9873,
       "step": 29600
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4699287170133254e-05,
-      "loss": 2.8947,
+      "epoch": 0.1,
+      "learning_rate": 4.507878908385942e-05,
+      "loss": 2.9926,
       "step": 29700
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4681439652187575e-05,
-      "loss": 2.8999,
+      "epoch": 0.1,
+      "learning_rate": 4.5062219350135044e-05,
+      "loss": 2.9889,
       "step": 29800
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4663592134241897e-05,
-      "loss": 2.8911,
+      "epoch": 0.1,
+      "learning_rate": 4.5045649616410665e-05,
+      "loss": 2.9975,
       "step": 29900
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.464574461629622e-05,
-      "loss": 2.8956,
+      "epoch": 0.1,
+      "learning_rate": 4.502907988268629e-05,
+      "loss": 2.9792,
       "step": 30000
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.462789709835053e-05,
-      "loss": 2.9037,
+      "epoch": 0.1,
+      "learning_rate": 4.501251014896191e-05,
+      "loss": 2.9954,
       "step": 30100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4610049580404853e-05,
-      "loss": 2.888,
+      "epoch": 0.1,
+      "learning_rate": 4.499594041523753e-05,
+      "loss": 2.988,
       "step": 30200
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4592202062459175e-05,
-      "loss": 2.8946,
+      "epoch": 0.1,
+      "learning_rate": 4.497937068151315e-05,
+      "loss": 2.9855,
       "step": 30300
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4574354544513496e-05,
-      "loss": 2.9023,
+      "epoch": 0.1,
+      "learning_rate": 4.4962800947788766e-05,
+      "loss": 2.9948,
       "step": 30400
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.455650702656782e-05,
-      "loss": 2.9015,
+      "epoch": 0.1,
+      "learning_rate": 4.4946231214064394e-05,
+      "loss": 2.9854,
       "step": 30500
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.453865950862214e-05,
-      "loss": 2.8981,
+      "epoch": 0.1,
+      "learning_rate": 4.4929661480340015e-05,
+      "loss": 2.9906,
       "step": 30600
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.452081199067646e-05,
-      "loss": 2.9012,
+      "epoch": 0.1,
+      "learning_rate": 4.491309174661563e-05,
+      "loss": 2.9817,
       "step": 30700
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.450296447273078e-05,
-      "loss": 2.9027,
+      "epoch": 0.1,
+      "learning_rate": 4.489652201289126e-05,
+      "loss": 2.986,
       "step": 30800
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.44851169547851e-05,
-      "loss": 2.8998,
+      "epoch": 0.1,
+      "learning_rate": 4.487995227916687e-05,
+      "loss": 2.9968,
       "step": 30900
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.446726943683942e-05,
-      "loss": 2.896,
+      "epoch": 0.1,
+      "learning_rate": 4.4863382545442495e-05,
+      "loss": 2.9827,
       "step": 31000
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4449421918893744e-05,
-      "loss": 2.8943,
+      "epoch": 0.1,
+      "learning_rate": 4.484681281171812e-05,
+      "loss": 2.9964,
       "step": 31100
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.443157440094806e-05,
-      "loss": 2.8959,
+      "epoch": 0.1,
+      "learning_rate": 4.483024307799374e-05,
+      "loss": 2.9892,
       "step": 31200
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.441372688300238e-05,
-      "loss": 2.8864,
+      "epoch": 0.1,
+      "learning_rate": 4.481367334426936e-05,
+      "loss": 2.9747,
       "step": 31300
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.43958793650567e-05,
-      "loss": 2.8969,
+      "epoch": 0.1,
+      "learning_rate": 4.479710361054498e-05,
+      "loss": 2.9837,
       "step": 31400
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.437803184711102e-05,
-      "loss": 2.8877,
+      "epoch": 0.1,
+      "learning_rate": 4.47805338768206e-05,
+      "loss": 2.9814,
       "step": 31500
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.4360184329165344e-05,
-      "loss": 2.888,
+      "epoch": 0.1,
+      "learning_rate": 4.476396414309622e-05,
+      "loss": 2.9854,
       "step": 31600
     },
     {
       "epoch": 0.11,
-      "learning_rate": 4.4342336811219665e-05,
-      "loss": 2.8799,
+      "learning_rate": 4.4747394409371845e-05,
+      "loss": 2.9701,
       "step": 31700
     },
     {
       "epoch": 0.11,
-      "learning_rate": 4.4324489293273986e-05,
-      "loss": 2.8901,
+      "learning_rate": 4.4730824675647466e-05,
+      "loss": 2.9821,
       "step": 31800
     },
     {
       "epoch": 0.11,
-      "learning_rate": 4.430664177532831e-05,
-      "loss": 2.8944,
+      "learning_rate": 4.471425494192308e-05,
+      "loss": 2.9924,
       "step": 31900
     },
     {
       "epoch": 0.11,
-      "learning_rate": 4.428879425738263e-05,
-      "loss": 2.8914,
+      "learning_rate": 4.469768520819871e-05,
+      "loss": 2.9904,
       "step": 32000
     },
     {
       "epoch": 0.11,
-      "learning_rate": 4.427094673943695e-05,
-      "loss": 2.8816,
+      "learning_rate": 4.468111547447433e-05,
+      "loss": 2.9996,
       "step": 32100
     },
     {
       "epoch": 0.11,
-      "learning_rate": 4.425309922149127e-05,
-      "loss": 2.8752,
+      "learning_rate": 4.4664545740749945e-05,
+      "loss": 2.9968,
       "step": 32200
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.4235251703545585e-05,
-      "loss": 2.8947,
+      "epoch": 0.11,
+      "learning_rate": 4.4647976007025574e-05,
+      "loss": 2.983,
       "step": 32300
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.421740418559991e-05,
-      "loss": 2.8882,
+      "epoch": 0.11,
+      "learning_rate": 4.463140627330119e-05,
+      "loss": 2.985,
       "step": 32400
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.419955666765423e-05,
-      "loss": 2.8851,
+      "epoch": 0.11,
+      "learning_rate": 4.461483653957681e-05,
+      "loss": 2.9903,
       "step": 32500
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.418170914970855e-05,
-      "loss": 2.8871,
+      "epoch": 0.11,
+      "learning_rate": 4.459826680585243e-05,
+      "loss": 2.9813,
       "step": 32600
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.416386163176287e-05,
-      "loss": 2.8999,
+      "epoch": 0.11,
+      "learning_rate": 4.458169707212805e-05,
+      "loss": 2.9977,
       "step": 32700
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.414601411381719e-05,
-      "loss": 2.8968,
+      "epoch": 0.11,
+      "learning_rate": 4.4565127338403674e-05,
+      "loss": 2.9993,
       "step": 32800
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.412816659587151e-05,
-      "loss": 2.8961,
+      "epoch": 0.11,
+      "learning_rate": 4.4548557604679296e-05,
+      "loss": 2.9881,
       "step": 32900
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.4110319077925834e-05,
-      "loss": 2.8843,
+      "epoch": 0.11,
+      "learning_rate": 4.453198787095492e-05,
+      "loss": 2.9856,
       "step": 33000
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.4092471559980155e-05,
-      "loss": 2.8937,
+      "epoch": 0.11,
+      "learning_rate": 4.451541813723054e-05,
+      "loss": 2.9814,
       "step": 33100
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.4074624042034476e-05,
-      "loss": 2.8943,
+      "epoch": 0.11,
+      "learning_rate": 4.449884840350616e-05,
+      "loss": 2.9919,
       "step": 33200
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.40567765240888e-05,
-      "loss": 2.8934,
+      "epoch": 0.11,
+      "learning_rate": 4.448227866978178e-05,
+      "loss": 2.9812,
       "step": 33300
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.403892900614312e-05,
-      "loss": 2.8827,
+      "epoch": 0.11,
+      "learning_rate": 4.4465708936057396e-05,
+      "loss": 2.9909,
       "step": 33400
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.402108148819744e-05,
-      "loss": 2.8842,
+      "epoch": 0.11,
+      "learning_rate": 4.4449139202333024e-05,
+      "loss": 2.9874,
       "step": 33500
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.400323397025176e-05,
-      "loss": 2.897,
+      "epoch": 0.11,
+      "learning_rate": 4.443256946860864e-05,
+      "loss": 2.9795,
       "step": 33600
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.398538645230608e-05,
-      "loss": 2.8934,
+      "epoch": 0.11,
+      "learning_rate": 4.441599973488426e-05,
+      "loss": 2.989,
       "step": 33700
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3967538934360404e-05,
-      "loss": 2.8872,
+      "epoch": 0.11,
+      "learning_rate": 4.439943000115988e-05,
+      "loss": 2.991,
       "step": 33800
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3949691416414725e-05,
-      "loss": 2.8913,
+      "epoch": 0.11,
+      "learning_rate": 4.4382860267435504e-05,
+      "loss": 2.993,
       "step": 33900
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3931843898469046e-05,
-      "loss": 2.8967,
+      "epoch": 0.11,
+      "learning_rate": 4.4366290533711125e-05,
+      "loss": 2.9769,
       "step": 34000
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.391399638052337e-05,
-      "loss": 2.8891,
+      "epoch": 0.11,
+      "learning_rate": 4.4349720799986746e-05,
+      "loss": 2.9919,
       "step": 34100
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.389614886257769e-05,
-      "loss": 2.8905,
+      "epoch": 0.11,
+      "learning_rate": 4.433315106626237e-05,
+      "loss": 2.9988,
       "step": 34200
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3878301344632e-05,
-      "loss": 2.8873,
+      "epoch": 0.11,
+      "learning_rate": 4.431658133253799e-05,
+      "loss": 2.9766,
       "step": 34300
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3860453826686324e-05,
-      "loss": 2.8666,
+      "epoch": 0.11,
+      "learning_rate": 4.4300011598813604e-05,
+      "loss": 2.9893,
       "step": 34400
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3842606308740645e-05,
-      "loss": 2.8838,
+      "epoch": 0.11,
+      "learning_rate": 4.428344186508923e-05,
+      "loss": 2.9957,
       "step": 34500
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.3824758790794967e-05,
-      "loss": 2.8856,
+      "epoch": 0.11,
+      "learning_rate": 4.4266872131364854e-05,
+      "loss": 2.9998,
       "step": 34600
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.380691127284929e-05,
-      "loss": 2.8823,
+      "epoch": 0.11,
+      "learning_rate": 4.425030239764047e-05,
+      "loss": 2.9961,
       "step": 34700
     },
     {
       "epoch": 0.12,
-      "learning_rate": 4.378906375490361e-05,
-      "loss": 2.8862,
+      "learning_rate": 4.42337326639161e-05,
+      "loss": 2.9779,
       "step": 34800
     },
     {
       "epoch": 0.12,
-      "learning_rate": 4.377121623695793e-05,
-      "loss": 2.8845,
+      "learning_rate": 4.421716293019171e-05,
+      "loss": 2.9903,
       "step": 34900
     },
     {
       "epoch": 0.12,
-      "learning_rate": 4.375336871901225e-05,
-      "loss": 2.8771,
+      "learning_rate": 4.420059319646733e-05,
+      "loss": 2.9791,
       "step": 35000
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.373552120106657e-05,
-      "loss": 2.8801,
+      "epoch": 0.12,
+      "learning_rate": 4.4184023462742954e-05,
+      "loss": 2.9834,
       "step": 35100
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3717673683120894e-05,
-      "loss": 2.8853,
+      "epoch": 0.12,
+      "learning_rate": 4.4167453729018576e-05,
+      "loss": 2.9692,
       "step": 35200
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.369982616517521e-05,
-      "loss": 2.8846,
+      "epoch": 0.12,
+      "learning_rate": 4.41508839952942e-05,
+      "loss": 2.9811,
       "step": 35300
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.368197864722953e-05,
-      "loss": 2.8738,
+      "epoch": 0.12,
+      "learning_rate": 4.413431426156982e-05,
+      "loss": 2.9805,
       "step": 35400
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.366413112928385e-05,
-      "loss": 2.8921,
+      "epoch": 0.12,
+      "learning_rate": 4.411774452784544e-05,
+      "loss": 2.9802,
       "step": 35500
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.364628361133817e-05,
-      "loss": 2.8795,
+      "epoch": 0.12,
+      "learning_rate": 4.410117479412106e-05,
+      "loss": 2.9779,
       "step": 35600
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.362843609339249e-05,
-      "loss": 2.8811,
+      "epoch": 0.12,
+      "learning_rate": 4.408460506039668e-05,
+      "loss": 2.9947,
       "step": 35700
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3610588575446814e-05,
-      "loss": 2.8788,
+      "epoch": 0.12,
+      "learning_rate": 4.4068035326672305e-05,
+      "loss": 2.9649,
       "step": 35800
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3592741057501136e-05,
-      "loss": 2.8811,
+      "epoch": 0.12,
+      "learning_rate": 4.405146559294792e-05,
+      "loss": 2.9809,
       "step": 35900
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.357489353955546e-05,
-      "loss": 2.8831,
+      "epoch": 0.12,
+      "learning_rate": 4.403489585922355e-05,
+      "loss": 2.9851,
       "step": 36000
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.355704602160978e-05,
-      "loss": 2.8804,
+      "epoch": 0.12,
+      "learning_rate": 4.401832612549916e-05,
+      "loss": 2.9785,
       "step": 36100
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.35391985036641e-05,
-      "loss": 2.8745,
+      "epoch": 0.12,
+      "learning_rate": 4.4001756391774784e-05,
+      "loss": 2.9807,
       "step": 36200
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.352135098571842e-05,
-      "loss": 2.8806,
+      "epoch": 0.12,
+      "learning_rate": 4.398518665805041e-05,
+      "loss": 2.9675,
       "step": 36300
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3503503467772735e-05,
-      "loss": 2.889,
+      "epoch": 0.12,
+      "learning_rate": 4.396861692432603e-05,
+      "loss": 2.9754,
       "step": 36400
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3485655949827056e-05,
-      "loss": 2.8757,
+      "epoch": 0.12,
+      "learning_rate": 4.395204719060165e-05,
+      "loss": 2.9817,
       "step": 36500
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.346780843188138e-05,
-      "loss": 2.8937,
+      "epoch": 0.12,
+      "learning_rate": 4.393547745687727e-05,
+      "loss": 2.9777,
       "step": 36600
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.34499609139357e-05,
-      "loss": 2.8649,
+      "epoch": 0.12,
+      "learning_rate": 4.391890772315289e-05,
+      "loss": 2.9775,
       "step": 36700
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.343211339599002e-05,
-      "loss": 2.8782,
+      "epoch": 0.12,
+      "learning_rate": 4.390233798942851e-05,
+      "loss": 2.9892,
       "step": 36800
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.341426587804434e-05,
-      "loss": 2.8843,
+      "epoch": 0.12,
+      "learning_rate": 4.3885768255704134e-05,
+      "loss": 2.9768,
       "step": 36900
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.339641836009866e-05,
-      "loss": 2.8823,
+      "epoch": 0.12,
+      "learning_rate": 4.3869198521979755e-05,
+      "loss": 2.971,
       "step": 37000
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3378570842152983e-05,
-      "loss": 2.8677,
+      "epoch": 0.12,
+      "learning_rate": 4.385262878825537e-05,
+      "loss": 2.9787,
       "step": 37100
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3360723324207305e-05,
-      "loss": 2.8788,
+      "epoch": 0.12,
+      "learning_rate": 4.3836059054531e-05,
+      "loss": 2.9678,
       "step": 37200
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3342875806261626e-05,
-      "loss": 2.8773,
+      "epoch": 0.12,
+      "learning_rate": 4.381948932080662e-05,
+      "loss": 2.9739,
       "step": 37300
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.332502828831595e-05,
-      "loss": 2.8774,
+      "epoch": 0.12,
+      "learning_rate": 4.3802919587082235e-05,
+      "loss": 2.9653,
       "step": 37400
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.330718077037026e-05,
-      "loss": 2.8756,
+      "epoch": 0.12,
+      "learning_rate": 4.378634985335786e-05,
+      "loss": 2.9768,
       "step": 37500
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.328933325242458e-05,
-      "loss": 2.8975,
+      "epoch": 0.12,
+      "learning_rate": 4.376978011963348e-05,
+      "loss": 2.9761,
       "step": 37600
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.3271485734478904e-05,
-      "loss": 2.8696,
+      "epoch": 0.12,
+      "learning_rate": 4.37532103859091e-05,
+      "loss": 2.9683,
       "step": 37700
     },
     {
       "epoch": 0.13,
-      "learning_rate": 4.3253638216533225e-05,
-      "loss": 2.8816,
+      "learning_rate": 4.373664065218473e-05,
+      "loss": 2.9778,
       "step": 37800
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.3235790698587546e-05,
-      "loss": 2.8705,
+      "epoch": 0.13,
+      "learning_rate": 4.372007091846034e-05,
+      "loss": 2.9728,
       "step": 37900
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.321794318064187e-05,
-      "loss": 2.8702,
+      "epoch": 0.13,
+      "learning_rate": 4.370350118473596e-05,
+      "loss": 2.9739,
       "step": 38000
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.320009566269619e-05,
-      "loss": 2.8791,
+      "epoch": 0.13,
+      "learning_rate": 4.3686931451011585e-05,
+      "loss": 2.9666,
       "step": 38100
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.318224814475051e-05,
-      "loss": 2.8812,
+      "epoch": 0.13,
+      "learning_rate": 4.3670361717287206e-05,
+      "loss": 2.9773,
       "step": 38200
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.316440062680484e-05,
-      "loss": 2.868,
+      "epoch": 0.13,
+      "learning_rate": 4.365379198356283e-05,
+      "loss": 2.9898,
       "step": 38300
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.314655310885915e-05,
-      "loss": 2.881,
+      "epoch": 0.13,
+      "learning_rate": 4.363722224983845e-05,
+      "loss": 2.9769,
       "step": 38400
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.3128705590913474e-05,
-      "loss": 2.8783,
+      "epoch": 0.13,
+      "learning_rate": 4.362065251611407e-05,
+      "loss": 2.9648,
       "step": 38500
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.3110858072967795e-05,
-      "loss": 2.8848,
+      "epoch": 0.13,
+      "learning_rate": 4.3604082782389685e-05,
+      "loss": 2.9713,
       "step": 38600
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.3093010555022116e-05,
-      "loss": 2.8771,
+      "epoch": 0.13,
+      "learning_rate": 4.3587513048665314e-05,
+      "loss": 2.9766,
       "step": 38700
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.307516303707644e-05,
-      "loss": 2.8584,
+      "epoch": 0.13,
+      "learning_rate": 4.3570943314940935e-05,
+      "loss": 2.9695,
       "step": 38800
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.305731551913076e-05,
-      "loss": 2.8824,
+      "epoch": 0.13,
+      "learning_rate": 4.355437358121655e-05,
+      "loss": 2.9651,
       "step": 38900
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.303946800118508e-05,
-      "loss": 2.879,
+      "epoch": 0.13,
+      "learning_rate": 4.353780384749217e-05,
+      "loss": 2.977,
       "step": 39000
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.30216204832394e-05,
-      "loss": 2.8753,
+      "epoch": 0.13,
+      "learning_rate": 4.352123411376779e-05,
+      "loss": 2.9685,
       "step": 39100
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.300377296529372e-05,
-      "loss": 2.8864,
+      "epoch": 0.13,
+      "learning_rate": 4.3504664380043414e-05,
+      "loss": 2.9731,
       "step": 39200
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.298592544734804e-05,
-      "loss": 2.8741,
+      "epoch": 0.13,
+      "learning_rate": 4.3488094646319036e-05,
+      "loss": 2.9733,
       "step": 39300
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2968077929402365e-05,
-      "loss": 2.8719,
+      "epoch": 0.13,
+      "learning_rate": 4.347152491259466e-05,
+      "loss": 2.9907,
       "step": 39400
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.295023041145668e-05,
-      "loss": 2.8661,
+      "epoch": 0.13,
+      "learning_rate": 4.345495517887028e-05,
+      "loss": 2.9816,
       "step": 39500
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2932382893511e-05,
-      "loss": 2.8768,
+      "epoch": 0.13,
+      "learning_rate": 4.343838544514589e-05,
+      "loss": 2.97,
       "step": 39600
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.291453537556532e-05,
-      "loss": 2.858,
+      "epoch": 0.13,
+      "learning_rate": 4.342181571142152e-05,
+      "loss": 2.9801,
       "step": 39700
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.289668785761964e-05,
-      "loss": 2.8816,
+      "epoch": 0.13,
+      "learning_rate": 4.340524597769714e-05,
+      "loss": 2.9809,
       "step": 39800
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2878840339673964e-05,
-      "loss": 2.8814,
+      "epoch": 0.13,
+      "learning_rate": 4.338867624397276e-05,
+      "loss": 2.9679,
       "step": 39900
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2860992821728285e-05,
-      "loss": 2.8632,
+      "epoch": 0.13,
+      "learning_rate": 4.3372106510248386e-05,
+      "loss": 2.9769,
       "step": 40000
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2843145303782606e-05,
-      "loss": 2.8693,
+      "epoch": 0.13,
+      "learning_rate": 4.3355536776524e-05,
+      "loss": 2.9735,
       "step": 40100
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.282529778583693e-05,
-      "loss": 2.8715,
+      "epoch": 0.13,
+      "learning_rate": 4.333896704279962e-05,
+      "loss": 2.9713,
       "step": 40200
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.280745026789125e-05,
-      "loss": 2.8737,
+      "epoch": 0.13,
+      "learning_rate": 4.3322397309075244e-05,
+      "loss": 2.97,
       "step": 40300
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.278960274994557e-05,
-      "loss": 2.8663,
+      "epoch": 0.13,
+      "learning_rate": 4.3305827575350865e-05,
+      "loss": 2.9733,
       "step": 40400
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.277175523199989e-05,
-      "loss": 2.8713,
+      "epoch": 0.13,
+      "learning_rate": 4.3289257841626486e-05,
+      "loss": 2.9825,
       "step": 40500
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.2753907714054206e-05,
-      "loss": 2.869,
+      "epoch": 0.13,
+      "learning_rate": 4.327268810790211e-05,
+      "loss": 2.9765,
       "step": 40600
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.273606019610853e-05,
-      "loss": 2.8638,
+      "epoch": 0.13,
+      "learning_rate": 4.325611837417773e-05,
+      "loss": 2.969,
       "step": 40700
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.271821267816285e-05,
-      "loss": 2.8783,
+      "epoch": 0.14,
+      "learning_rate": 4.323954864045335e-05,
+      "loss": 2.9672,
       "step": 40800
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.270036516021717e-05,
-      "loss": 2.8704,
+      "epoch": 0.14,
+      "learning_rate": 4.322297890672897e-05,
+      "loss": 2.9695,
       "step": 40900
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.268251764227149e-05,
-      "loss": 2.8665,
+      "epoch": 0.14,
+      "learning_rate": 4.3206409173004594e-05,
+      "loss": 2.9889,
       "step": 41000
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.266467012432581e-05,
-      "loss": 2.8649,
+      "epoch": 0.14,
+      "learning_rate": 4.318983943928021e-05,
+      "loss": 2.9629,
       "step": 41100
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.264682260638013e-05,
-      "loss": 2.8722,
+      "epoch": 0.14,
+      "learning_rate": 4.317326970555584e-05,
+      "loss": 2.9674,
       "step": 41200
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2628975088434454e-05,
-      "loss": 2.8696,
+      "epoch": 0.14,
+      "learning_rate": 4.315669997183145e-05,
+      "loss": 2.9775,
       "step": 41300
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2611127570488775e-05,
-      "loss": 2.8669,
+      "epoch": 0.14,
+      "learning_rate": 4.314013023810707e-05,
+      "loss": 2.9828,
       "step": 41400
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2593280052543097e-05,
-      "loss": 2.8631,
+      "epoch": 0.14,
+      "learning_rate": 4.31235605043827e-05,
+      "loss": 2.9738,
       "step": 41500
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.257543253459742e-05,
-      "loss": 2.8778,
+      "epoch": 0.14,
+      "learning_rate": 4.3106990770658316e-05,
+      "loss": 2.9856,
       "step": 41600
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.255758501665173e-05,
-      "loss": 2.8643,
+      "epoch": 0.14,
+      "learning_rate": 4.309042103693394e-05,
+      "loss": 2.9638,
       "step": 41700
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2539737498706053e-05,
-      "loss": 2.854,
+      "epoch": 0.14,
+      "learning_rate": 4.307385130320956e-05,
+      "loss": 2.9762,
       "step": 41800
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2521889980760375e-05,
-      "loss": 2.8628,
+      "epoch": 0.14,
+      "learning_rate": 4.305728156948518e-05,
+      "loss": 2.9594,
       "step": 41900
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2504042462814696e-05,
-      "loss": 2.8594,
+      "epoch": 0.14,
+      "learning_rate": 4.30407118357608e-05,
+      "loss": 2.9796,
       "step": 42000
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.248619494486902e-05,
-      "loss": 2.8626,
+      "epoch": 0.14,
+      "learning_rate": 4.302414210203642e-05,
+      "loss": 2.976,
       "step": 42100
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.246834742692334e-05,
-      "loss": 2.8617,
+      "epoch": 0.14,
+      "learning_rate": 4.3007572368312045e-05,
+      "loss": 2.9723,
       "step": 42200
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.245049990897766e-05,
-      "loss": 2.8608,
+      "epoch": 0.14,
+      "learning_rate": 4.299100263458766e-05,
+      "loss": 2.9761,
       "step": 42300
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.243265239103198e-05,
-      "loss": 2.8686,
+      "epoch": 0.14,
+      "learning_rate": 4.297443290086329e-05,
+      "loss": 2.9771,
       "step": 42400
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.24148048730863e-05,
-      "loss": 2.8715,
+      "epoch": 0.14,
+      "learning_rate": 4.295786316713891e-05,
+      "loss": 2.9687,
       "step": 42500
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.239695735514062e-05,
-      "loss": 2.8743,
+      "epoch": 0.14,
+      "learning_rate": 4.2941293433414524e-05,
+      "loss": 2.9633,
       "step": 42600
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2379109837194944e-05,
-      "loss": 2.8715,
+      "epoch": 0.14,
+      "learning_rate": 4.292472369969015e-05,
+      "loss": 2.9761,
       "step": 42700
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.236126231924926e-05,
-      "loss": 2.8652,
+      "epoch": 0.14,
+      "learning_rate": 4.290815396596577e-05,
+      "loss": 2.966,
       "step": 42800
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.234341480130358e-05,
-      "loss": 2.8489,
+      "epoch": 0.14,
+      "learning_rate": 4.289158423224139e-05,
+      "loss": 2.9734,
       "step": 42900
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.23255672833579e-05,
-      "loss": 2.8605,
+      "epoch": 0.14,
+      "learning_rate": 4.2875014498517016e-05,
+      "loss": 2.9552,
       "step": 43000
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.230771976541222e-05,
-      "loss": 2.8606,
+      "epoch": 0.14,
+      "learning_rate": 4.285844476479263e-05,
+      "loss": 2.9892,
       "step": 43100
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2289872247466544e-05,
-      "loss": 2.8656,
+      "epoch": 0.14,
+      "learning_rate": 4.284187503106825e-05,
+      "loss": 2.969,
       "step": 43200
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2272024729520865e-05,
-      "loss": 2.8607,
+      "epoch": 0.14,
+      "learning_rate": 4.2825305297343874e-05,
+      "loss": 2.9798,
       "step": 43300
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.2254177211575186e-05,
-      "loss": 2.8614,
+      "epoch": 0.14,
+      "learning_rate": 4.2808735563619495e-05,
+      "loss": 2.9845,
       "step": 43400
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.2236329693629514e-05,
-      "loss": 2.858,
+      "epoch": 0.14,
+      "learning_rate": 4.279216582989512e-05,
+      "loss": 2.9577,
       "step": 43500
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.2218482175683835e-05,
-      "loss": 2.8582,
+      "epoch": 0.14,
+      "learning_rate": 4.277559609617074e-05,
+      "loss": 2.9702,
       "step": 43600
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.220063465773815e-05,
-      "loss": 2.8664,
+      "epoch": 0.14,
+      "learning_rate": 4.275902636244636e-05,
+      "loss": 2.9714,
       "step": 43700
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.218278713979247e-05,
-      "loss": 2.867,
+      "epoch": 0.15,
+      "learning_rate": 4.2742456628721975e-05,
+      "loss": 2.9726,
       "step": 43800
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.216493962184679e-05,
-      "loss": 2.8731,
+      "epoch": 0.15,
+      "learning_rate": 4.27258868949976e-05,
+      "loss": 2.97,
       "step": 43900
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.214709210390111e-05,
-      "loss": 2.8612,
+      "epoch": 0.15,
+      "learning_rate": 4.2709317161273224e-05,
+      "loss": 2.9742,
       "step": 44000
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.2129244585955435e-05,
-      "loss": 2.8553,
+      "epoch": 0.15,
+      "learning_rate": 4.269274742754884e-05,
+      "loss": 2.9739,
       "step": 44100
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.2111397068009756e-05,
-      "loss": 2.8642,
+      "epoch": 0.15,
+      "learning_rate": 4.267617769382446e-05,
+      "loss": 2.9667,
       "step": 44200
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.209354955006408e-05,
-      "loss": 2.8693,
+      "epoch": 0.15,
+      "learning_rate": 4.265960796010008e-05,
+      "loss": 2.9496,
       "step": 44300
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.20757020321184e-05,
-      "loss": 2.8573,
+      "epoch": 0.15,
+      "learning_rate": 4.26430382263757e-05,
+      "loss": 2.9808,
       "step": 44400
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.205785451417272e-05,
-      "loss": 2.8646,
+      "epoch": 0.15,
+      "learning_rate": 4.2626468492651325e-05,
+      "loss": 2.9846,
       "step": 44500
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.204000699622704e-05,
-      "loss": 2.863,
+      "epoch": 0.15,
+      "learning_rate": 4.2609898758926946e-05,
+      "loss": 2.9769,
       "step": 44600
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.202215947828136e-05,
-      "loss": 2.8674,
+      "epoch": 0.15,
+      "learning_rate": 4.259332902520257e-05,
+      "loss": 2.9771,
       "step": 44700
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.2004311960335676e-05,
-      "loss": 2.8609,
+      "epoch": 0.15,
+      "learning_rate": 4.257675929147818e-05,
+      "loss": 2.9718,
       "step": 44800
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.198646444239e-05,
-      "loss": 2.8604,
+      "epoch": 0.15,
+      "learning_rate": 4.256018955775381e-05,
+      "loss": 2.9591,
       "step": 44900
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.196861692444432e-05,
-      "loss": 2.8587,
+      "epoch": 0.15,
+      "learning_rate": 4.254361982402943e-05,
+      "loss": 2.9632,
       "step": 45000
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.195076940649864e-05,
-      "loss": 2.8611,
+      "epoch": 0.15,
+      "learning_rate": 4.252705009030505e-05,
+      "loss": 2.9734,
       "step": 45100
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.193292188855296e-05,
-      "loss": 2.8605,
+      "epoch": 0.15,
+      "learning_rate": 4.2510480356580675e-05,
+      "loss": 2.9647,
       "step": 45200
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.191507437060728e-05,
-      "loss": 2.8599,
+      "epoch": 0.15,
+      "learning_rate": 4.249391062285629e-05,
+      "loss": 2.9547,
       "step": 45300
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.1897226852661604e-05,
-      "loss": 2.8597,
+      "epoch": 0.15,
+      "learning_rate": 4.247734088913191e-05,
+      "loss": 2.971,
       "step": 45400
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.1879379334715925e-05,
-      "loss": 2.8613,
+      "epoch": 0.15,
+      "learning_rate": 4.246077115540753e-05,
+      "loss": 2.964,
       "step": 45500
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.1861531816770246e-05,
-      "loss": 2.8523,
+      "epoch": 0.15,
+      "learning_rate": 4.2444201421683154e-05,
+      "loss": 2.9686,
       "step": 45600
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.184368429882457e-05,
-      "loss": 2.8463,
+      "epoch": 0.15,
+      "learning_rate": 4.2427631687958776e-05,
+      "loss": 2.9567,
       "step": 45700
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.182583678087889e-05,
-      "loss": 2.8558,
+      "epoch": 0.15,
+      "learning_rate": 4.24110619542344e-05,
+      "loss": 2.9584,
       "step": 45800
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.18079892629332e-05,
-      "loss": 2.854,
+      "epoch": 0.15,
+      "learning_rate": 4.239449222051002e-05,
+      "loss": 2.9728,
       "step": 45900
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.1790141744987524e-05,
-      "loss": 2.8539,
+      "epoch": 0.15,
+      "learning_rate": 4.237792248678564e-05,
+      "loss": 2.9764,
       "step": 46000
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.1772294227041845e-05,
-      "loss": 2.8612,
+      "epoch": 0.15,
+      "learning_rate": 4.236135275306126e-05,
+      "loss": 2.9648,
       "step": 46100
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.1754446709096167e-05,
-      "loss": 2.8571,
+      "epoch": 0.15,
+      "learning_rate": 4.234478301933688e-05,
+      "loss": 2.9742,
       "step": 46200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.173659919115049e-05,
-      "loss": 2.8656,
+      "epoch": 0.15,
+      "learning_rate": 4.23282132856125e-05,
+      "loss": 2.9651,
       "step": 46300
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.171875167320481e-05,
-      "loss": 2.8571,
+      "epoch": 0.15,
+      "learning_rate": 4.2311643551888126e-05,
+      "loss": 2.9748,
       "step": 46400
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.170090415525913e-05,
-      "loss": 2.8538,
+      "epoch": 0.15,
+      "learning_rate": 4.229507381816375e-05,
+      "loss": 2.9504,
       "step": 46500
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.168305663731345e-05,
-      "loss": 2.8643,
+      "epoch": 0.15,
+      "learning_rate": 4.227850408443936e-05,
+      "loss": 2.9823,
       "step": 46600
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.166520911936777e-05,
-      "loss": 2.8593,
+      "epoch": 0.15,
+      "learning_rate": 4.226193435071499e-05,
+      "loss": 2.9503,
       "step": 46700
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.1647361601422094e-05,
-      "loss": 2.8608,
+      "epoch": 0.16,
+      "learning_rate": 4.2245364616990605e-05,
+      "loss": 2.9667,
       "step": 46800
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.1629514083476415e-05,
-      "loss": 2.8656,
+      "epoch": 0.16,
+      "learning_rate": 4.2228794883266226e-05,
+      "loss": 2.9621,
       "step": 46900
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.161166656553073e-05,
-      "loss": 2.8556,
+      "epoch": 0.16,
+      "learning_rate": 4.221222514954185e-05,
+      "loss": 2.9614,
       "step": 47000
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.159381904758505e-05,
-      "loss": 2.8533,
+      "epoch": 0.16,
+      "learning_rate": 4.219565541581747e-05,
+      "loss": 2.9631,
       "step": 47100
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.157597152963937e-05,
-      "loss": 2.8573,
+      "epoch": 0.16,
+      "learning_rate": 4.217908568209309e-05,
+      "loss": 2.9569,
       "step": 47200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.155812401169369e-05,
-      "loss": 2.8553,
+      "epoch": 0.16,
+      "learning_rate": 4.216251594836871e-05,
+      "loss": 2.9594,
       "step": 47300
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.1540276493748014e-05,
-      "loss": 2.8608,
+      "epoch": 0.16,
+      "learning_rate": 4.2145946214644334e-05,
+      "loss": 2.9717,
       "step": 47400
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.1522428975802336e-05,
-      "loss": 2.8571,
+      "epoch": 0.16,
+      "learning_rate": 4.2129376480919955e-05,
+      "loss": 2.9775,
       "step": 47500
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.150458145785666e-05,
-      "loss": 2.8527,
+      "epoch": 0.16,
+      "learning_rate": 4.211280674719558e-05,
+      "loss": 2.9613,
       "step": 47600
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.148673393991098e-05,
-      "loss": 2.8564,
+      "epoch": 0.16,
+      "learning_rate": 4.20962370134712e-05,
+      "loss": 2.9595,
       "step": 47700
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.14688864219653e-05,
-      "loss": 2.8488,
+      "epoch": 0.16,
+      "learning_rate": 4.207966727974681e-05,
+      "loss": 2.9608,
       "step": 47800
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.145103890401962e-05,
-      "loss": 2.8546,
+      "epoch": 0.16,
+      "learning_rate": 4.206309754602244e-05,
+      "loss": 2.97,
       "step": 47900
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.143319138607394e-05,
-      "loss": 2.8542,
+      "epoch": 0.16,
+      "learning_rate": 4.2046527812298056e-05,
+      "loss": 2.9739,
       "step": 48000
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.1415343868128256e-05,
-      "loss": 2.8474,
+      "epoch": 0.16,
+      "learning_rate": 4.202995807857368e-05,
+      "loss": 2.9706,
       "step": 48100
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.139749635018258e-05,
-      "loss": 2.8504,
+      "epoch": 0.16,
+      "learning_rate": 4.2013388344849306e-05,
+      "loss": 2.9679,
       "step": 48200
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.13796488322369e-05,
-      "loss": 2.858,
+      "epoch": 0.16,
+      "learning_rate": 4.199681861112492e-05,
+      "loss": 2.9685,
       "step": 48300
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.136180131429122e-05,
-      "loss": 2.847,
+      "epoch": 0.16,
+      "learning_rate": 4.198024887740054e-05,
+      "loss": 2.9605,
       "step": 48400
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.134395379634554e-05,
-      "loss": 2.8431,
+      "epoch": 0.16,
+      "learning_rate": 4.196367914367616e-05,
+      "loss": 2.9572,
       "step": 48500
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.132610627839987e-05,
-      "loss": 2.8534,
+      "epoch": 0.16,
+      "learning_rate": 4.1947109409951785e-05,
+      "loss": 2.9714,
       "step": 48600
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.130825876045419e-05,
-      "loss": 2.8457,
+      "epoch": 0.16,
+      "learning_rate": 4.1930539676227406e-05,
+      "loss": 2.9644,
       "step": 48700
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.129041124250851e-05,
-      "loss": 2.8611,
+      "epoch": 0.16,
+      "learning_rate": 4.191396994250303e-05,
+      "loss": 2.9603,
       "step": 48800
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.127256372456283e-05,
-      "loss": 2.8424,
+      "epoch": 0.16,
+      "learning_rate": 4.189740020877865e-05,
+      "loss": 2.9734,
       "step": 48900
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.125471620661715e-05,
-      "loss": 2.8518,
+      "epoch": 0.16,
+      "learning_rate": 4.1880830475054264e-05,
+      "loss": 2.9657,
       "step": 49000
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.123686868867147e-05,
-      "loss": 2.8601,
+      "epoch": 0.16,
+      "learning_rate": 4.186426074132989e-05,
+      "loss": 2.9651,
       "step": 49100
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.121902117072579e-05,
-      "loss": 2.8475,
+      "epoch": 0.16,
+      "learning_rate": 4.1847691007605513e-05,
+      "loss": 2.9695,
       "step": 49200
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.120117365278011e-05,
-      "loss": 2.8561,
+      "epoch": 0.16,
+      "learning_rate": 4.183112127388113e-05,
+      "loss": 2.9548,
       "step": 49300
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.118332613483443e-05,
-      "loss": 2.8666,
+      "epoch": 0.16,
+      "learning_rate": 4.181455154015675e-05,
+      "loss": 2.9679,
       "step": 49400
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.116547861688875e-05,
-      "loss": 2.8522,
+      "epoch": 0.16,
+      "learning_rate": 4.179798180643237e-05,
+      "loss": 2.9611,
       "step": 49500
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1147631098943074e-05,
-      "loss": 2.8481,
+      "epoch": 0.16,
+      "learning_rate": 4.178141207270799e-05,
+      "loss": 2.9584,
       "step": 49600
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1129783580997395e-05,
-      "loss": 2.8584,
+      "epoch": 0.16,
+      "learning_rate": 4.1764842338983614e-05,
+      "loss": 2.9616,
       "step": 49700
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.111193606305172e-05,
-      "loss": 2.8557,
+      "epoch": 0.17,
+      "learning_rate": 4.1748272605259235e-05,
+      "loss": 2.9627,
       "step": 49800
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.109408854510604e-05,
-      "loss": 2.8509,
+      "epoch": 0.17,
+      "learning_rate": 4.173170287153486e-05,
+      "loss": 2.9463,
       "step": 49900
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.107624102716036e-05,
-      "loss": 2.8536,
+      "epoch": 0.17,
+      "learning_rate": 4.171513313781047e-05,
+      "loss": 2.9457,
       "step": 50000
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1058393509214674e-05,
-      "loss": 2.843,
+      "epoch": 0.17,
+      "learning_rate": 4.16985634040861e-05,
+      "loss": 2.9502,
       "step": 50100
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1040545991268995e-05,
-      "loss": 2.848,
+      "epoch": 0.17,
+      "learning_rate": 4.168199367036172e-05,
+      "loss": 2.9453,
       "step": 50200
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.1022698473323316e-05,
-      "loss": 2.8521,
+      "epoch": 0.17,
+      "learning_rate": 4.1665423936637336e-05,
+      "loss": 2.95,
       "step": 50300
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.100485095537764e-05,
-      "loss": 2.8368,
+      "epoch": 0.17,
+      "learning_rate": 4.1648854202912964e-05,
+      "loss": 2.9536,
       "step": 50400
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.098700343743196e-05,
-      "loss": 2.8653,
+      "epoch": 0.17,
+      "learning_rate": 4.163228446918858e-05,
+      "loss": 2.976,
       "step": 50500
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.096915591948628e-05,
-      "loss": 2.8587,
+      "epoch": 0.17,
+      "learning_rate": 4.16157147354642e-05,
+      "loss": 2.9517,
       "step": 50600
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.09513084015406e-05,
-      "loss": 2.8547,
+      "epoch": 0.17,
+      "learning_rate": 4.159914500173983e-05,
+      "loss": 2.9522,
       "step": 50700
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.093346088359492e-05,
-      "loss": 2.847,
+      "epoch": 0.17,
+      "learning_rate": 4.158257526801544e-05,
+      "loss": 2.9652,
       "step": 50800
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.091561336564924e-05,
-      "loss": 2.8457,
+      "epoch": 0.17,
+      "learning_rate": 4.1566005534291065e-05,
+      "loss": 2.9542,
       "step": 50900
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.0897765847703565e-05,
-      "loss": 2.8517,
+      "epoch": 0.17,
+      "learning_rate": 4.1549435800566686e-05,
+      "loss": 2.9668,
       "step": 51000
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.0879918329757886e-05,
-      "loss": 2.8362,
+      "epoch": 0.17,
+      "learning_rate": 4.153286606684231e-05,
+      "loss": 2.9634,
       "step": 51100
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.08620708118122e-05,
-      "loss": 2.8562,
+      "epoch": 0.17,
+      "learning_rate": 4.151629633311793e-05,
+      "loss": 2.9566,
       "step": 51200
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.084422329386652e-05,
-      "loss": 2.8518,
+      "epoch": 0.17,
+      "learning_rate": 4.149972659939355e-05,
+      "loss": 2.9573,
       "step": 51300
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.082637577592084e-05,
-      "loss": 2.8371,
+      "epoch": 0.17,
+      "learning_rate": 4.148315686566917e-05,
+      "loss": 2.9608,
       "step": 51400
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.0808528257975164e-05,
-      "loss": 2.8484,
+      "epoch": 0.17,
+      "learning_rate": 4.146658713194479e-05,
+      "loss": 2.9746,
       "step": 51500
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.0790680740029485e-05,
-      "loss": 2.8457,
+      "epoch": 0.17,
+      "learning_rate": 4.1450017398220415e-05,
+      "loss": 2.96,
       "step": 51600
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.0772833222083806e-05,
-      "loss": 2.843,
+      "epoch": 0.17,
+      "learning_rate": 4.1433447664496037e-05,
+      "loss": 2.9583,
       "step": 51700
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.075498570413813e-05,
-      "loss": 2.8378,
+      "epoch": 0.17,
+      "learning_rate": 4.141687793077165e-05,
+      "loss": 2.9679,
       "step": 51800
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.073713818619245e-05,
-      "loss": 2.8577,
+      "epoch": 0.17,
+      "learning_rate": 4.140030819704728e-05,
+      "loss": 2.9531,
       "step": 51900
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.071929066824677e-05,
-      "loss": 2.8458,
+      "epoch": 0.17,
+      "learning_rate": 4.1383738463322894e-05,
+      "loss": 2.9548,
       "step": 52000
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.070144315030109e-05,
-      "loss": 2.8408,
+      "epoch": 0.17,
+      "learning_rate": 4.1367168729598516e-05,
+      "loss": 2.9615,
       "step": 52100
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.068359563235541e-05,
-      "loss": 2.8462,
+      "epoch": 0.17,
+      "learning_rate": 4.135059899587414e-05,
+      "loss": 2.9516,
       "step": 52200
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.066574811440973e-05,
-      "loss": 2.828,
+      "epoch": 0.17,
+      "learning_rate": 4.133402926214976e-05,
+      "loss": 2.9498,
       "step": 52300
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.064790059646405e-05,
-      "loss": 2.8387,
+      "epoch": 0.17,
+      "learning_rate": 4.131745952842538e-05,
+      "loss": 2.9666,
       "step": 52400
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.063005307851837e-05,
-      "loss": 2.8412,
+      "epoch": 0.17,
+      "learning_rate": 4.1300889794701e-05,
+      "loss": 2.9536,
       "step": 52500
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.061220556057269e-05,
-      "loss": 2.8409,
+      "epoch": 0.17,
+      "learning_rate": 4.128432006097662e-05,
+      "loss": 2.9581,
       "step": 52600
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.059435804262701e-05,
-      "loss": 2.8331,
+      "epoch": 0.17,
+      "learning_rate": 4.1267750327252244e-05,
+      "loss": 2.9615,
       "step": 52700
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.057651052468133e-05,
-      "loss": 2.8533,
+      "epoch": 0.17,
+      "learning_rate": 4.1251180593527866e-05,
+      "loss": 2.9539,
       "step": 52800
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0558663006735654e-05,
-      "loss": 2.8419,
+      "epoch": 0.18,
+      "learning_rate": 4.123461085980349e-05,
+      "loss": 2.9649,
       "step": 52900
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0540815488789975e-05,
-      "loss": 2.844,
+      "epoch": 0.18,
+      "learning_rate": 4.12180411260791e-05,
+      "loss": 2.9482,
       "step": 53000
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0522967970844296e-05,
-      "loss": 2.8462,
+      "epoch": 0.18,
+      "learning_rate": 4.120147139235473e-05,
+      "loss": 2.9532,
       "step": 53100
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.050512045289862e-05,
-      "loss": 2.8403,
+      "epoch": 0.18,
+      "learning_rate": 4.1184901658630345e-05,
+      "loss": 2.9598,
       "step": 53200
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.048727293495294e-05,
-      "loss": 2.8335,
+      "epoch": 0.18,
+      "learning_rate": 4.1168331924905966e-05,
+      "loss": 2.9606,
       "step": 53300
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.046942541700725e-05,
-      "loss": 2.8462,
+      "epoch": 0.18,
+      "learning_rate": 4.1151762191181595e-05,
+      "loss": 2.9687,
       "step": 53400
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0451577899061575e-05,
-      "loss": 2.838,
+      "epoch": 0.18,
+      "learning_rate": 4.113519245745721e-05,
+      "loss": 2.9533,
       "step": 53500
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0433730381115896e-05,
-      "loss": 2.8352,
+      "epoch": 0.18,
+      "learning_rate": 4.111862272373283e-05,
+      "loss": 2.9577,
       "step": 53600
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0415882863170224e-05,
-      "loss": 2.8374,
+      "epoch": 0.18,
+      "learning_rate": 4.110205299000845e-05,
+      "loss": 2.9455,
       "step": 53700
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0398035345224545e-05,
-      "loss": 2.8347,
+      "epoch": 0.18,
+      "learning_rate": 4.1085483256284074e-05,
+      "loss": 2.9502,
       "step": 53800
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0380187827278866e-05,
-      "loss": 2.8214,
+      "epoch": 0.18,
+      "learning_rate": 4.1068913522559695e-05,
+      "loss": 2.9447,
       "step": 53900
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.036234030933319e-05,
-      "loss": 2.849,
+      "epoch": 0.18,
+      "learning_rate": 4.105234378883532e-05,
+      "loss": 2.9683,
       "step": 54000
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.034449279138751e-05,
-      "loss": 2.8423,
+      "epoch": 0.18,
+      "learning_rate": 4.103577405511094e-05,
+      "loss": 2.9701,
       "step": 54100
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.032664527344183e-05,
-      "loss": 2.8396,
+      "epoch": 0.18,
+      "learning_rate": 4.101920432138655e-05,
+      "loss": 2.9478,
       "step": 54200
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0308797755496144e-05,
-      "loss": 2.8333,
+      "epoch": 0.18,
+      "learning_rate": 4.100263458766218e-05,
+      "loss": 2.9494,
       "step": 54300
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.0290950237550465e-05,
-      "loss": 2.8334,
+      "epoch": 0.18,
+      "learning_rate": 4.09860648539378e-05,
+      "loss": 2.9568,
       "step": 54400
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.027310271960479e-05,
-      "loss": 2.8289,
+      "epoch": 0.18,
+      "learning_rate": 4.096949512021342e-05,
+      "loss": 2.9516,
       "step": 54500
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.025525520165911e-05,
-      "loss": 2.8505,
+      "epoch": 0.18,
+      "learning_rate": 4.095292538648904e-05,
+      "loss": 2.9514,
       "step": 54600
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.023740768371343e-05,
-      "loss": 2.8387,
+      "epoch": 0.18,
+      "learning_rate": 4.093635565276466e-05,
+      "loss": 2.9522,
       "step": 54700
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.021956016576775e-05,
-      "loss": 2.8407,
+      "epoch": 0.18,
+      "learning_rate": 4.091978591904028e-05,
+      "loss": 2.9713,
       "step": 54800
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.020171264782207e-05,
-      "loss": 2.8396,
+      "epoch": 0.18,
+      "learning_rate": 4.09032161853159e-05,
+      "loss": 2.957,
       "step": 54900
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.018386512987639e-05,
-      "loss": 2.8391,
+      "epoch": 0.18,
+      "learning_rate": 4.0886646451591525e-05,
+      "loss": 2.9486,
       "step": 55000
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.0166017611930714e-05,
-      "loss": 2.8314,
+      "epoch": 0.18,
+      "learning_rate": 4.0870076717867146e-05,
+      "loss": 2.9382,
       "step": 55100
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.0148170093985035e-05,
-      "loss": 2.8403,
+      "epoch": 0.18,
+      "learning_rate": 4.085350698414277e-05,
+      "loss": 2.9393,
       "step": 55200
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.0130322576039356e-05,
-      "loss": 2.842,
+      "epoch": 0.18,
+      "learning_rate": 4.083693725041839e-05,
+      "loss": 2.9436,
       "step": 55300
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.011247505809367e-05,
-      "loss": 2.8382,
+      "epoch": 0.18,
+      "learning_rate": 4.082036751669401e-05,
+      "loss": 2.9463,
       "step": 55400
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.009462754014799e-05,
-      "loss": 2.8445,
+      "epoch": 0.18,
+      "learning_rate": 4.0803797782969625e-05,
+      "loss": 2.9397,
       "step": 55500
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.007678002220231e-05,
-      "loss": 2.8377,
+      "epoch": 0.18,
+      "learning_rate": 4.0787228049245253e-05,
+      "loss": 2.9508,
       "step": 55600
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.0058932504256634e-05,
-      "loss": 2.8388,
+      "epoch": 0.18,
+      "learning_rate": 4.077065831552087e-05,
+      "loss": 2.9523,
       "step": 55700
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.0041084986310956e-05,
-      "loss": 2.8286,
+      "epoch": 0.18,
+      "learning_rate": 4.075408858179649e-05,
+      "loss": 2.9636,
       "step": 55800
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.002323746836528e-05,
-      "loss": 2.8204,
+      "epoch": 0.19,
+      "learning_rate": 4.073751884807212e-05,
+      "loss": 2.9482,
       "step": 55900
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.00053899504196e-05,
-      "loss": 2.8434,
+      "epoch": 0.19,
+      "learning_rate": 4.072094911434773e-05,
+      "loss": 2.9608,
       "step": 56000
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.998754243247392e-05,
-      "loss": 2.8375,
+      "epoch": 0.19,
+      "learning_rate": 4.0704379380623354e-05,
+      "loss": 2.9604,
       "step": 56100
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.996969491452824e-05,
-      "loss": 2.8283,
+      "epoch": 0.19,
+      "learning_rate": 4.0687809646898975e-05,
+      "loss": 2.9551,
       "step": 56200
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.995184739658256e-05,
-      "loss": 2.8268,
+      "epoch": 0.19,
+      "learning_rate": 4.06712399131746e-05,
+      "loss": 2.9558,
       "step": 56300
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.993399987863688e-05,
-      "loss": 2.8338,
+      "epoch": 0.19,
+      "learning_rate": 4.065467017945022e-05,
+      "loss": 2.9506,
       "step": 56400
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.99161523606912e-05,
-      "loss": 2.8414,
+      "epoch": 0.19,
+      "learning_rate": 4.063810044572584e-05,
+      "loss": 2.9501,
       "step": 56500
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.989830484274552e-05,
-      "loss": 2.8335,
+      "epoch": 0.19,
+      "learning_rate": 4.062153071200146e-05,
+      "loss": 2.9625,
       "step": 56600
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.988045732479984e-05,
-      "loss": 2.8301,
+      "epoch": 0.19,
+      "learning_rate": 4.0604960978277076e-05,
+      "loss": 2.9422,
       "step": 56700
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.986260980685416e-05,
-      "loss": 2.8353,
+      "epoch": 0.19,
+      "learning_rate": 4.0588391244552704e-05,
+      "loss": 2.9616,
       "step": 56800
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.984476228890848e-05,
-      "loss": 2.8319,
+      "epoch": 0.19,
+      "learning_rate": 4.0571821510828326e-05,
+      "loss": 2.9585,
       "step": 56900
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.9826914770962804e-05,
-      "loss": 2.8368,
+      "epoch": 0.19,
+      "learning_rate": 4.055525177710394e-05,
+      "loss": 2.9564,
       "step": 57000
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.9809067253017125e-05,
-      "loss": 2.8305,
+      "epoch": 0.19,
+      "learning_rate": 4.053868204337957e-05,
+      "loss": 2.9594,
       "step": 57100
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.9791219735071446e-05,
-      "loss": 2.8269,
+      "epoch": 0.19,
+      "learning_rate": 4.052211230965518e-05,
+      "loss": 2.9484,
       "step": 57200
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.977337221712577e-05,
-      "loss": 2.8326,
+      "epoch": 0.19,
+      "learning_rate": 4.0505542575930805e-05,
+      "loss": 2.9537,
       "step": 57300
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 3.975552469918009e-05,
-      "loss": 2.8347,
+      "epoch": 0.19,
+      "learning_rate": 4.0488972842206426e-05,
+      "loss": 2.9594,
       "step": 57400
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.97376771812344e-05,
-      "loss": 2.8247,
+      "epoch": 0.19,
+      "learning_rate": 4.047240310848205e-05,
+      "loss": 2.9532,
       "step": 57500
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9719829663288724e-05,
-      "loss": 2.8293,
+      "epoch": 0.19,
+      "learning_rate": 4.045583337475767e-05,
+      "loss": 2.9452,
       "step": 57600
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9701982145343045e-05,
-      "loss": 2.8288,
+      "epoch": 0.19,
+      "learning_rate": 4.043926364103329e-05,
+      "loss": 2.9582,
       "step": 57700
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9684134627397366e-05,
-      "loss": 2.8195,
+      "epoch": 0.19,
+      "learning_rate": 4.042269390730891e-05,
+      "loss": 2.9602,
       "step": 57800
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.966628710945169e-05,
-      "loss": 2.8307,
+      "epoch": 0.19,
+      "learning_rate": 4.0406124173584534e-05,
+      "loss": 2.9563,
       "step": 57900
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.964843959150601e-05,
-      "loss": 2.8316,
+      "epoch": 0.19,
+      "learning_rate": 4.0389554439860155e-05,
+      "loss": 2.9529,
       "step": 58000
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.963059207356033e-05,
-      "loss": 2.835,
+      "epoch": 0.19,
+      "learning_rate": 4.0372984706135777e-05,
+      "loss": 2.9622,
       "step": 58100
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.961274455561465e-05,
-      "loss": 2.8219,
+      "epoch": 0.19,
+      "learning_rate": 4.035641497241139e-05,
+      "loss": 2.9484,
       "step": 58200
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.959489703766897e-05,
-      "loss": 2.8291,
+      "epoch": 0.19,
+      "learning_rate": 4.033984523868702e-05,
+      "loss": 2.9722,
       "step": 58300
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9577049519723294e-05,
-      "loss": 2.8261,
+      "epoch": 0.19,
+      "learning_rate": 4.032327550496264e-05,
+      "loss": 2.9554,
       "step": 58400
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9559202001777615e-05,
-      "loss": 2.8232,
+      "epoch": 0.19,
+      "learning_rate": 4.0306705771238256e-05,
+      "loss": 2.9504,
       "step": 58500
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.954135448383193e-05,
-      "loss": 2.8356,
+      "epoch": 0.19,
+      "learning_rate": 4.0290136037513884e-05,
+      "loss": 2.9631,
       "step": 58600
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.952350696588625e-05,
-      "loss": 2.8329,
+      "epoch": 0.19,
+      "learning_rate": 4.02735663037895e-05,
+      "loss": 2.9529,
       "step": 58700
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.950565944794058e-05,
-      "loss": 2.8351,
+      "epoch": 0.19,
+      "learning_rate": 4.025699657006512e-05,
+      "loss": 2.9439,
       "step": 58800
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.94878119299949e-05,
-      "loss": 2.8262,
+      "epoch": 0.2,
+      "learning_rate": 4.024042683634074e-05,
+      "loss": 2.9489,
       "step": 58900
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.946996441204922e-05,
-      "loss": 2.8217,
+      "epoch": 0.2,
+      "learning_rate": 4.022385710261636e-05,
+      "loss": 2.9402,
       "step": 59000
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.945211689410354e-05,
-      "loss": 2.8293,
+      "epoch": 0.2,
+      "learning_rate": 4.0207287368891984e-05,
+      "loss": 2.948,
       "step": 59100
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9434269376157863e-05,
-      "loss": 2.8246,
+      "epoch": 0.2,
+      "learning_rate": 4.0190717635167606e-05,
+      "loss": 2.9486,
       "step": 59200
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9416421858212185e-05,
-      "loss": 2.8263,
+      "epoch": 0.2,
+      "learning_rate": 4.017414790144323e-05,
+      "loss": 2.9589,
       "step": 59300
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9398574340266506e-05,
-      "loss": 2.8327,
+      "epoch": 0.2,
+      "learning_rate": 4.015757816771885e-05,
+      "loss": 2.945,
       "step": 59400
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.938072682232082e-05,
-      "loss": 2.8302,
+      "epoch": 0.2,
+      "learning_rate": 4.014100843399447e-05,
+      "loss": 2.9483,
       "step": 59500
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.936287930437514e-05,
-      "loss": 2.8261,
+      "epoch": 0.2,
+      "learning_rate": 4.012443870027009e-05,
+      "loss": 2.9519,
       "step": 59600
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.934503178642946e-05,
-      "loss": 2.8255,
+      "epoch": 0.2,
+      "learning_rate": 4.0107868966545707e-05,
+      "loss": 2.9483,
       "step": 59700
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9327184268483784e-05,
-      "loss": 2.8291,
+      "epoch": 0.2,
+      "learning_rate": 4.009129923282133e-05,
+      "loss": 2.9495,
       "step": 59800
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9309336750538105e-05,
-      "loss": 2.8335,
+      "epoch": 0.2,
+      "learning_rate": 4.007472949909695e-05,
+      "loss": 2.9585,
       "step": 59900
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.9291489232592426e-05,
-      "loss": 2.8345,
+      "epoch": 0.2,
+      "learning_rate": 4.005815976537257e-05,
+      "loss": 2.9392,
       "step": 60000
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.927364171464675e-05,
-      "loss": 2.8267,
+      "epoch": 0.2,
+      "learning_rate": 4.004159003164819e-05,
+      "loss": 2.948,
       "step": 60100
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 3.925579419670107e-05,
-      "loss": 2.8294,
+      "epoch": 0.2,
+      "learning_rate": 4.0025020297923814e-05,
+      "loss": 2.9519,
       "step": 60200
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.923794667875539e-05,
-      "loss": 2.8296,
+      "epoch": 0.2,
+      "learning_rate": 4.0008450564199435e-05,
+      "loss": 2.9558,
       "step": 60300
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.922009916080971e-05,
-      "loss": 2.825,
+      "epoch": 0.2,
+      "learning_rate": 3.999188083047506e-05,
+      "loss": 2.9527,
       "step": 60400
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.920225164286403e-05,
-      "loss": 2.8255,
+      "epoch": 0.2,
+      "learning_rate": 3.997531109675068e-05,
+      "loss": 2.9525,
       "step": 60500
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.918440412491835e-05,
-      "loss": 2.8265,
+      "epoch": 0.2,
+      "learning_rate": 3.99587413630263e-05,
+      "loss": 2.9464,
       "step": 60600
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.916655660697267e-05,
-      "loss": 2.824,
+      "epoch": 0.2,
+      "learning_rate": 3.9942171629301914e-05,
+      "loss": 2.9464,
       "step": 60700
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.914870908902699e-05,
-      "loss": 2.8239,
+      "epoch": 0.2,
+      "learning_rate": 3.992560189557754e-05,
+      "loss": 2.9417,
       "step": 60800
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.913086157108131e-05,
-      "loss": 2.8292,
+      "epoch": 0.2,
+      "learning_rate": 3.990903216185316e-05,
+      "loss": 2.9461,
       "step": 60900
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.911301405313563e-05,
-      "loss": 2.8268,
+      "epoch": 0.2,
+      "learning_rate": 3.989246242812878e-05,
+      "loss": 2.9593,
       "step": 61000
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.909516653518995e-05,
-      "loss": 2.8234,
+      "epoch": 0.2,
+      "learning_rate": 3.987589269440441e-05,
+      "loss": 2.9494,
       "step": 61100
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.9077319017244274e-05,
-      "loss": 2.8276,
+      "epoch": 0.2,
+      "learning_rate": 3.985932296068002e-05,
+      "loss": 2.9514,
       "step": 61200
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.9059471499298595e-05,
-      "loss": 2.8208,
+      "epoch": 0.2,
+      "learning_rate": 3.984275322695564e-05,
+      "loss": 2.9408,
       "step": 61300
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.904162398135292e-05,
-      "loss": 2.8301,
+      "epoch": 0.2,
+      "learning_rate": 3.9826183493231265e-05,
+      "loss": 2.9581,
       "step": 61400
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.902377646340724e-05,
-      "loss": 2.8209,
+      "epoch": 0.2,
+      "learning_rate": 3.9809613759506886e-05,
+      "loss": 2.9452,
       "step": 61500
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.900592894546156e-05,
-      "loss": 2.8219,
+      "epoch": 0.2,
+      "learning_rate": 3.979304402578251e-05,
+      "loss": 2.941,
       "step": 61600
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8988081427515874e-05,
-      "loss": 2.8197,
+      "epoch": 0.2,
+      "learning_rate": 3.977647429205813e-05,
+      "loss": 2.9409,
       "step": 61700
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8970233909570195e-05,
-      "loss": 2.828,
+      "epoch": 0.2,
+      "learning_rate": 3.975990455833375e-05,
+      "loss": 2.9514,
       "step": 61800
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8952386391624516e-05,
-      "loss": 2.8193,
+      "epoch": 0.21,
+      "learning_rate": 3.9743334824609365e-05,
+      "loss": 2.9499,
       "step": 61900
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.893453887367884e-05,
-      "loss": 2.824,
+      "epoch": 0.21,
+      "learning_rate": 3.9726765090884993e-05,
+      "loss": 2.9306,
       "step": 62000
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.891669135573316e-05,
-      "loss": 2.8243,
+      "epoch": 0.21,
+      "learning_rate": 3.9710195357160615e-05,
+      "loss": 2.9382,
       "step": 62100
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.889884383778748e-05,
-      "loss": 2.8251,
+      "epoch": 0.21,
+      "learning_rate": 3.969362562343623e-05,
+      "loss": 2.9546,
       "step": 62200
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.88809963198418e-05,
-      "loss": 2.8178,
+      "epoch": 0.21,
+      "learning_rate": 3.967705588971186e-05,
+      "loss": 2.9356,
       "step": 62300
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.886314880189612e-05,
-      "loss": 2.8137,
+      "epoch": 0.21,
+      "learning_rate": 3.966048615598747e-05,
+      "loss": 2.9409,
       "step": 62400
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.884530128395044e-05,
-      "loss": 2.8303,
+      "epoch": 0.21,
+      "learning_rate": 3.9643916422263094e-05,
+      "loss": 2.9496,
       "step": 62500
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8827453766004764e-05,
-      "loss": 2.8176,
+      "epoch": 0.21,
+      "learning_rate": 3.962734668853872e-05,
+      "loss": 2.9528,
       "step": 62600
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.8809606248059086e-05,
-      "loss": 2.8195,
+      "epoch": 0.21,
+      "learning_rate": 3.961077695481434e-05,
+      "loss": 2.9499,
       "step": 62700
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.87917587301134e-05,
-      "loss": 2.8333,
+      "epoch": 0.21,
+      "learning_rate": 3.959420722108996e-05,
+      "loss": 2.9493,
       "step": 62800
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.877391121216772e-05,
-      "loss": 2.8282,
+      "epoch": 0.21,
+      "learning_rate": 3.957763748736558e-05,
+      "loss": 2.9554,
       "step": 62900
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 3.875606369422204e-05,
-      "loss": 2.8149,
+      "epoch": 0.21,
+      "learning_rate": 3.95610677536412e-05,
+      "loss": 2.9564,
       "step": 63000
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.8738216176276364e-05,
-      "loss": 2.8329,
+      "epoch": 0.21,
+      "learning_rate": 3.954449801991682e-05,
+      "loss": 2.9339,
       "step": 63100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.8720368658330685e-05,
-      "loss": 2.8254,
+      "epoch": 0.21,
+      "learning_rate": 3.9527928286192444e-05,
+      "loss": 2.9401,
       "step": 63200
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.8702521140385006e-05,
-      "loss": 2.8217,
+      "epoch": 0.21,
+      "learning_rate": 3.9511358552468066e-05,
+      "loss": 2.9466,
       "step": 63300
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.868467362243933e-05,
-      "loss": 2.8281,
+      "epoch": 0.21,
+      "learning_rate": 3.949478881874368e-05,
+      "loss": 2.9408,
       "step": 63400
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.866682610449365e-05,
-      "loss": 2.8274,
+      "epoch": 0.21,
+      "learning_rate": 3.947821908501931e-05,
+      "loss": 2.9416,
       "step": 63500
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.864897858654797e-05,
-      "loss": 2.8173,
+      "epoch": 0.21,
+      "learning_rate": 3.946164935129493e-05,
+      "loss": 2.9482,
       "step": 63600
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.863113106860229e-05,
-      "loss": 2.8199,
+      "epoch": 0.21,
+      "learning_rate": 3.9445079617570545e-05,
+      "loss": 2.949,
       "step": 63700
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.861328355065661e-05,
-      "loss": 2.8304,
+      "epoch": 0.21,
+      "learning_rate": 3.942850988384617e-05,
+      "loss": 2.9546,
       "step": 63800
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.8595436032710933e-05,
-      "loss": 2.8193,
+      "epoch": 0.21,
+      "learning_rate": 3.941194015012179e-05,
+      "loss": 2.9471,
       "step": 63900
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.8577588514765255e-05,
-      "loss": 2.8193,
+      "epoch": 0.21,
+      "learning_rate": 3.939537041639741e-05,
+      "loss": 2.9347,
       "step": 64000
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.8559740996819576e-05,
-      "loss": 2.8161,
+      "epoch": 0.21,
+      "learning_rate": 3.937880068267303e-05,
+      "loss": 2.9509,
       "step": 64100
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.85418934788739e-05,
-      "loss": 2.822,
+      "epoch": 0.21,
+      "learning_rate": 3.936223094894865e-05,
+      "loss": 2.9609,
       "step": 64200
     },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9345661215224274e-05,
+      "loss": 2.9484,
+      "step": 64300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9329091481499895e-05,
+      "loss": 2.9575,
+      "step": 64400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9312521747775517e-05,
+      "loss": 2.9475,
+      "step": 64500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.929595201405114e-05,
+      "loss": 2.946,
+      "step": 64600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.927938228032676e-05,
+      "loss": 2.9413,
+      "step": 64700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.926281254660238e-05,
+      "loss": 2.9431,
+      "step": 64800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9246242812877996e-05,
+      "loss": 2.9392,
+      "step": 64900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.922967307915362e-05,
+      "loss": 2.9446,
+      "step": 65000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.921310334542924e-05,
+      "loss": 2.9548,
+      "step": 65100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.919653361170486e-05,
+      "loss": 2.946,
+      "step": 65200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.917996387798048e-05,
+      "loss": 2.9623,
+      "step": 65300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91633941442561e-05,
+      "loss": 2.9381,
+      "step": 65400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9146824410531724e-05,
+      "loss": 2.9438,
+      "step": 65500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9130254676807346e-05,
+      "loss": 2.9537,
+      "step": 65600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.911368494308297e-05,
+      "loss": 2.9364,
+      "step": 65700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.909711520935859e-05,
+      "loss": 2.9341,
+      "step": 65800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9080545475634204e-05,
+      "loss": 2.9341,
+      "step": 65900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.906397574190983e-05,
+      "loss": 2.9431,
+      "step": 66000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9047406008185447e-05,
+      "loss": 2.939,
+      "step": 66100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.903083627446107e-05,
+      "loss": 2.9492,
+      "step": 66200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9014266540736696e-05,
+      "loss": 2.9485,
+      "step": 66300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.899769680701231e-05,
+      "loss": 2.9418,
+      "step": 66400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.898112707328793e-05,
+      "loss": 2.9534,
+      "step": 66500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8964557339563554e-05,
+      "loss": 2.9379,
+      "step": 66600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8947987605839175e-05,
+      "loss": 2.951,
+      "step": 66700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.89314178721148e-05,
+      "loss": 2.9286,
+      "step": 66800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.891484813839042e-05,
+      "loss": 2.9522,
+      "step": 66900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.889827840466604e-05,
+      "loss": 2.9438,
+      "step": 67000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.888170867094166e-05,
+      "loss": 2.9332,
+      "step": 67100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.886513893721728e-05,
+      "loss": 2.9434,
+      "step": 67200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8848569203492904e-05,
+      "loss": 2.9487,
+      "step": 67300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.883199946976852e-05,
+      "loss": 2.9458,
+      "step": 67400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.881542973604415e-05,
+      "loss": 2.9296,
+      "step": 67500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.879886000231976e-05,
+      "loss": 2.9305,
+      "step": 67600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.878229026859538e-05,
+      "loss": 2.9355,
+      "step": 67700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.876572053487101e-05,
+      "loss": 2.9476,
+      "step": 67800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8749150801146626e-05,
+      "loss": 2.9354,
+      "step": 67900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.873258106742225e-05,
+      "loss": 2.9301,
+      "step": 68000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.871601133369787e-05,
+      "loss": 2.9245,
+      "step": 68100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.869944159997349e-05,
+      "loss": 2.9275,
+      "step": 68200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.868287186624911e-05,
+      "loss": 2.9456,
+      "step": 68300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8666302132524733e-05,
+      "loss": 2.9429,
+      "step": 68400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8649732398800355e-05,
+      "loss": 2.949,
+      "step": 68500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.863316266507597e-05,
+      "loss": 2.9333,
+      "step": 68600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.86165929313516e-05,
+      "loss": 2.9455,
+      "step": 68700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.860002319762722e-05,
+      "loss": 2.938,
+      "step": 68800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8583453463902834e-05,
+      "loss": 2.9454,
+      "step": 68900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.856688373017846e-05,
+      "loss": 2.9369,
+      "step": 69000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.855031399645408e-05,
+      "loss": 2.9463,
+      "step": 69100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85337442627297e-05,
+      "loss": 2.9353,
+      "step": 69200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.851717452900532e-05,
+      "loss": 2.9361,
+      "step": 69300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.850060479528094e-05,
+      "loss": 2.9289,
+      "step": 69400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.848403506155656e-05,
+      "loss": 2.9493,
+      "step": 69500
+    },
     {
       "epoch": 0.23,
-      "learning_rate": 3.852404596092822e-05,
-      "loss": 2.8195,
-      "step": 64300
+      "learning_rate": 3.8467465327832184e-05,
+      "loss": 2.9466,
+      "step": 69600
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.850619844298254e-05,
-      "loss": 2.836,
-      "step": 64400
+      "learning_rate": 3.8450895594107806e-05,
+      "loss": 2.9454,
+      "step": 69700
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.848835092503686e-05,
-      "loss": 2.8261,
-      "step": 64500
+      "learning_rate": 3.843432586038343e-05,
+      "loss": 2.9391,
+      "step": 69800
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.847050340709118e-05,
-      "loss": 2.8232,
-      "step": 64600
+      "learning_rate": 3.841775612665905e-05,
+      "loss": 2.9463,
+      "step": 69900
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.84526558891455e-05,
-      "loss": 2.8256,
-      "step": 64700
+      "learning_rate": 3.840118639293467e-05,
+      "loss": 2.9571,
+      "step": 70000
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.843480837119982e-05,
-      "loss": 2.8224,
-      "step": 64800
+      "learning_rate": 3.8384616659210285e-05,
+      "loss": 2.9495,
+      "step": 70100
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.841696085325414e-05,
-      "loss": 2.8168,
-      "step": 64900
+      "learning_rate": 3.8368046925485906e-05,
+      "loss": 2.9483,
+      "step": 70200
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.839911333530846e-05,
-      "loss": 2.8229,
-      "step": 65000
+      "learning_rate": 3.8351477191761535e-05,
+      "loss": 2.9353,
+      "step": 70300
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.838126581736278e-05,
-      "loss": 2.8128,
-      "step": 65100
+      "learning_rate": 3.833490745803715e-05,
+      "loss": 2.9436,
+      "step": 70400
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.83634182994171e-05,
-      "loss": 2.8177,
-      "step": 65200
+      "learning_rate": 3.831833772431277e-05,
+      "loss": 2.9313,
+      "step": 70500
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.8345570781471424e-05,
-      "loss": 2.8243,
-      "step": 65300
+      "learning_rate": 3.830176799058839e-05,
+      "loss": 2.9506,
+      "step": 70600
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.8327723263525745e-05,
-      "loss": 2.8253,
-      "step": 65400
+      "learning_rate": 3.8285198256864014e-05,
+      "loss": 2.9537,
+      "step": 70700
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.8309875745580066e-05,
-      "loss": 2.8213,
-      "step": 65500
+      "learning_rate": 3.8268628523139635e-05,
+      "loss": 2.9269,
+      "step": 70800
     },
     {
       "epoch": 0.23,
-      "learning_rate": 3.829202822763439e-05,
-      "loss": 2.8196,
-      "step": 65600
+      "learning_rate": 3.8252058789415257e-05,
+      "loss": 2.9342,
+      "step": 70900
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.827418070968871e-05,
-      "loss": 2.8226,
-      "step": 65700
+      "epoch": 0.24,
+      "learning_rate": 3.823548905569088e-05,
+      "loss": 2.9363,
+      "step": 71000
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 3.825633319174303e-05,
-      "loss": 2.8216,
-      "step": 65800
+      "epoch": 0.24,
+      "learning_rate": 3.821891932196649e-05,
+      "loss": 2.9342,
+      "step": 71100
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.8238485673797344e-05,
-      "loss": 2.8176,
-      "step": 65900
+      "learning_rate": 3.820234958824212e-05,
+      "loss": 2.9365,
+      "step": 71200
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.8220638155851665e-05,
-      "loss": 2.8155,
-      "step": 66000
+      "learning_rate": 3.818577985451774e-05,
+      "loss": 2.929,
+      "step": 71300
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.820279063790599e-05,
-      "loss": 2.8205,
-      "step": 66100
+      "learning_rate": 3.816921012079336e-05,
+      "loss": 2.9437,
+      "step": 71400
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.818494311996031e-05,
-      "loss": 2.8173,
-      "step": 66200
+      "learning_rate": 3.8152640387068985e-05,
+      "loss": 2.9256,
+      "step": 71500
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.816709560201463e-05,
-      "loss": 2.8177,
-      "step": 66300
+      "learning_rate": 3.81360706533446e-05,
+      "loss": 2.9349,
+      "step": 71600
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.814924808406895e-05,
-      "loss": 2.8228,
-      "step": 66400
+      "learning_rate": 3.811950091962022e-05,
+      "loss": 2.9512,
+      "step": 71700
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.813140056612327e-05,
-      "loss": 2.8189,
-      "step": 66500
+      "learning_rate": 3.810293118589584e-05,
+      "loss": 2.942,
+      "step": 71800
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.811355304817759e-05,
-      "loss": 2.8131,
-      "step": 66600
+      "learning_rate": 3.8086361452171464e-05,
+      "loss": 2.9315,
+      "step": 71900
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.8095705530231914e-05,
-      "loss": 2.8143,
-      "step": 66700
+      "learning_rate": 3.8069791718447086e-05,
+      "loss": 2.937,
+      "step": 72000
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.8077858012286235e-05,
-      "loss": 2.8232,
-      "step": 66800
+      "learning_rate": 3.805322198472271e-05,
+      "loss": 2.9347,
+      "step": 72100
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.8060010494340556e-05,
-      "loss": 2.8274,
-      "step": 66900
+      "learning_rate": 3.803665225099833e-05,
+      "loss": 2.9404,
+      "step": 72200
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.804216297639487e-05,
-      "loss": 2.8239,
-      "step": 67000
+      "learning_rate": 3.802008251727395e-05,
+      "loss": 2.933,
+      "step": 72300
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.802431545844919e-05,
-      "loss": 2.8201,
-      "step": 67100
+      "learning_rate": 3.800351278354957e-05,
+      "loss": 2.9402,
+      "step": 72400
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.800646794050351e-05,
-      "loss": 2.8165,
-      "step": 67200
+      "learning_rate": 3.798694304982519e-05,
+      "loss": 2.9381,
+      "step": 72500
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.7988620422557834e-05,
-      "loss": 2.8003,
-      "step": 67300
+      "learning_rate": 3.797037331610081e-05,
+      "loss": 2.9295,
+      "step": 72600
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.7970772904612156e-05,
-      "loss": 2.8125,
-      "step": 67400
+      "learning_rate": 3.7953803582376436e-05,
+      "loss": 2.9467,
+      "step": 72700
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.795292538666648e-05,
-      "loss": 2.8208,
-      "step": 67500
+      "learning_rate": 3.793723384865205e-05,
+      "loss": 2.9189,
+      "step": 72800
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.79350778687208e-05,
-      "loss": 2.8136,
-      "step": 67600
+      "learning_rate": 3.792066411492767e-05,
+      "loss": 2.9247,
+      "step": 72900
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.791723035077512e-05,
-      "loss": 2.815,
-      "step": 67700
+      "learning_rate": 3.79040943812033e-05,
+      "loss": 2.9269,
+      "step": 73000
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.789938283282944e-05,
-      "loss": 2.8129,
-      "step": 67800
+      "learning_rate": 3.7887524647478915e-05,
+      "loss": 2.9443,
+      "step": 73100
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.788153531488376e-05,
-      "loss": 2.8263,
-      "step": 67900
+      "learning_rate": 3.787095491375454e-05,
+      "loss": 2.9276,
+      "step": 73200
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.786368779693808e-05,
-      "loss": 2.8188,
-      "step": 68000
+      "learning_rate": 3.785438518003016e-05,
+      "loss": 2.9255,
+      "step": 73300
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.78458402789924e-05,
-      "loss": 2.8157,
-      "step": 68100
+      "learning_rate": 3.783781544630578e-05,
+      "loss": 2.9458,
+      "step": 73400
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.782799276104672e-05,
-      "loss": 2.8161,
-      "step": 68200
+      "learning_rate": 3.78212457125814e-05,
+      "loss": 2.9304,
+      "step": 73500
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.781014524310104e-05,
-      "loss": 2.8165,
-      "step": 68300
+      "learning_rate": 3.780467597885702e-05,
+      "loss": 2.9414,
+      "step": 73600
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.779229772515536e-05,
-      "loss": 2.8123,
-      "step": 68400
+      "learning_rate": 3.7788106245132644e-05,
+      "loss": 2.9315,
+      "step": 73700
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.777445020720968e-05,
-      "loss": 2.8163,
-      "step": 68500
+      "learning_rate": 3.777153651140826e-05,
+      "loss": 2.9252,
+      "step": 73800
     },
     {
       "epoch": 0.24,
-      "learning_rate": 3.7756602689264003e-05,
-      "loss": 2.8181,
-      "step": 68600
+      "learning_rate": 3.775496677768389e-05,
+      "loss": 2.9382,
+      "step": 73900
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7738755171318325e-05,
-      "loss": 2.8145,
-      "step": 68700
+      "learning_rate": 3.773839704395951e-05,
+      "loss": 2.9316,
+      "step": 74000
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7720907653372646e-05,
-      "loss": 2.8149,
-      "step": 68800
+      "learning_rate": 3.772182731023512e-05,
+      "loss": 2.93,
+      "step": 74100
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.770306013542697e-05,
-      "loss": 2.811,
-      "step": 68900
+      "learning_rate": 3.770525757651075e-05,
+      "loss": 2.9336,
+      "step": 74200
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.768521261748129e-05,
-      "loss": 2.8121,
-      "step": 69000
+      "learning_rate": 3.7688687842786366e-05,
+      "loss": 2.9251,
+      "step": 74300
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.766736509953561e-05,
-      "loss": 2.8145,
-      "step": 69100
+      "learning_rate": 3.767211810906199e-05,
+      "loss": 2.939,
+      "step": 74400
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.764951758158993e-05,
-      "loss": 2.8179,
-      "step": 69200
+      "learning_rate": 3.7655548375337616e-05,
+      "loss": 2.9315,
+      "step": 74500
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.763167006364425e-05,
-      "loss": 2.8092,
-      "step": 69300
+      "learning_rate": 3.763897864161323e-05,
+      "loss": 2.9388,
+      "step": 74600
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.761382254569857e-05,
-      "loss": 2.8153,
-      "step": 69400
+      "learning_rate": 3.762240890788885e-05,
+      "loss": 2.9352,
+      "step": 74700
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7595975027752894e-05,
-      "loss": 2.8191,
-      "step": 69500
+      "learning_rate": 3.7605839174164473e-05,
+      "loss": 2.9284,
+      "step": 74800
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7578127509807216e-05,
-      "loss": 2.8241,
-      "step": 69600
+      "learning_rate": 3.7589269440440095e-05,
+      "loss": 2.9257,
+      "step": 74900
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.756027999186154e-05,
-      "loss": 2.812,
-      "step": 69700
+      "learning_rate": 3.7572699706715716e-05,
+      "loss": 2.9309,
+      "step": 75000
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.754243247391586e-05,
-      "loss": 2.8149,
-      "step": 69800
+      "learning_rate": 3.755612997299134e-05,
+      "loss": 2.9401,
+      "step": 75100
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.752458495597018e-05,
-      "loss": 2.8231,
-      "step": 69900
+      "learning_rate": 3.753956023926696e-05,
+      "loss": 2.9326,
+      "step": 75200
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.75067374380245e-05,
-      "loss": 2.8172,
-      "step": 70000
+      "learning_rate": 3.7522990505542574e-05,
+      "loss": 2.933,
+      "step": 75300
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7488889920078815e-05,
-      "loss": 2.8039,
-      "step": 70100
+      "learning_rate": 3.7506420771818196e-05,
+      "loss": 2.9301,
+      "step": 75400
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7471042402133136e-05,
-      "loss": 2.8142,
-      "step": 70200
+      "learning_rate": 3.7489851038093824e-05,
+      "loss": 2.9371,
+      "step": 75500
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.745319488418746e-05,
-      "loss": 2.8196,
-      "step": 70300
+      "learning_rate": 3.747328130436944e-05,
+      "loss": 2.9292,
+      "step": 75600
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.743534736624178e-05,
-      "loss": 2.8097,
-      "step": 70400
+      "learning_rate": 3.745671157064506e-05,
+      "loss": 2.9355,
+      "step": 75700
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.74174998482961e-05,
-      "loss": 2.8031,
-      "step": 70500
+      "learning_rate": 3.744014183692068e-05,
+      "loss": 2.9368,
+      "step": 75800
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.739965233035042e-05,
-      "loss": 2.8111,
-      "step": 70600
+      "learning_rate": 3.74235721031963e-05,
+      "loss": 2.9299,
+      "step": 75900
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.738180481240474e-05,
-      "loss": 2.8112,
-      "step": 70700
+      "learning_rate": 3.7407002369471924e-05,
+      "loss": 2.9394,
+      "step": 76000
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7363957294459063e-05,
-      "loss": 2.818,
-      "step": 70800
+      "learning_rate": 3.7390432635747546e-05,
+      "loss": 2.9193,
+      "step": 76100
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7346109776513385e-05,
-      "loss": 2.8037,
-      "step": 70900
+      "learning_rate": 3.737386290202317e-05,
+      "loss": 2.9254,
+      "step": 76200
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7328262258567706e-05,
-      "loss": 2.8099,
-      "step": 71000
+      "learning_rate": 3.735729316829878e-05,
+      "loss": 2.9286,
+      "step": 76300
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.731041474062203e-05,
-      "loss": 2.8041,
-      "step": 71100
+      "learning_rate": 3.734072343457441e-05,
+      "loss": 2.9371,
+      "step": 76400
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.729256722267634e-05,
-      "loss": 2.8205,
-      "step": 71200
+      "learning_rate": 3.732415370085003e-05,
+      "loss": 2.9345,
+      "step": 76500
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.727471970473066e-05,
-      "loss": 2.8189,
-      "step": 71300
+      "learning_rate": 3.7307583967125646e-05,
+      "loss": 2.9344,
+      "step": 76600
     },
     {
       "epoch": 0.25,
-      "learning_rate": 3.7256872186784984e-05,
-      "loss": 2.8132,
-      "step": 71400
+      "learning_rate": 3.7291014233401275e-05,
+      "loss": 2.9349,
+      "step": 76700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.727444449967689e-05,
+      "loss": 2.9303,
+      "step": 76800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.725787476595251e-05,
+      "loss": 2.9312,
+      "step": 76900
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.7239024668839305e-05,
-      "loss": 2.8115,
-      "step": 71500
+      "learning_rate": 3.724130503222813e-05,
+      "loss": 2.9384,
+      "step": 77000
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.7221177150893626e-05,
-      "loss": 2.8013,
-      "step": 71600
+      "learning_rate": 3.7224735298503754e-05,
+      "loss": 2.9199,
+      "step": 77100
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.720332963294795e-05,
-      "loss": 2.8026,
-      "step": 71700
+      "learning_rate": 3.7208165564779375e-05,
+      "loss": 2.9349,
+      "step": 77200
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.718548211500227e-05,
-      "loss": 2.8047,
-      "step": 71800
+      "learning_rate": 3.7191595831054997e-05,
+      "loss": 2.9206,
+      "step": 77300
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.716763459705659e-05,
-      "loss": 2.8157,
-      "step": 71900
+      "learning_rate": 3.717502609733062e-05,
+      "loss": 2.9406,
+      "step": 77400
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.714978707911091e-05,
-      "loss": 2.814,
-      "step": 72000
+      "learning_rate": 3.715845636360624e-05,
+      "loss": 2.9182,
+      "step": 77500
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.713193956116523e-05,
-      "loss": 2.8161,
-      "step": 72100
+      "learning_rate": 3.714188662988186e-05,
+      "loss": 2.9283,
+      "step": 77600
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.7114092043219554e-05,
-      "loss": 2.8115,
-      "step": 72200
+      "learning_rate": 3.712531689615748e-05,
+      "loss": 2.9316,
+      "step": 77700
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.709624452527387e-05,
-      "loss": 2.8193,
-      "step": 72300
+      "learning_rate": 3.71087471624331e-05,
+      "loss": 2.9298,
+      "step": 77800
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.707839700732819e-05,
-      "loss": 2.8088,
-      "step": 72400
+      "learning_rate": 3.7092177428708725e-05,
+      "loss": 2.9391,
+      "step": 77900
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.706054948938251e-05,
-      "loss": 2.804,
-      "step": 72500
+      "learning_rate": 3.707560769498434e-05,
+      "loss": 2.925,
+      "step": 78000
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.704270197143683e-05,
-      "loss": 2.798,
-      "step": 72600
+      "learning_rate": 3.705903796125996e-05,
+      "loss": 2.9325,
+      "step": 78100
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.702485445349115e-05,
-      "loss": 2.8055,
-      "step": 72700
+      "learning_rate": 3.704246822753559e-05,
+      "loss": 2.9292,
+      "step": 78200
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.7007006935545474e-05,
-      "loss": 2.8116,
-      "step": 72800
+      "learning_rate": 3.7025898493811205e-05,
+      "loss": 2.9319,
+      "step": 78300
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.6989159417599795e-05,
-      "loss": 2.8163,
-      "step": 72900
+      "learning_rate": 3.7009328760086826e-05,
+      "loss": 2.9372,
+      "step": 78400
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.6971311899654117e-05,
-      "loss": 2.8096,
-      "step": 73000
+      "learning_rate": 3.699275902636245e-05,
+      "loss": 2.9379,
+      "step": 78500
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.695346438170844e-05,
-      "loss": 2.8119,
-      "step": 73100
+      "learning_rate": 3.697618929263807e-05,
+      "loss": 2.918,
+      "step": 78600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.695961955891369e-05,
+      "loss": 2.9473,
+      "step": 78700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.694304982518931e-05,
+      "loss": 2.936,
+      "step": 78800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.692648009146493e-05,
+      "loss": 2.9358,
+      "step": 78900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6909910357740555e-05,
+      "loss": 2.9258,
+      "step": 79000
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.693561686376276e-05,
-      "loss": 2.8193,
-      "step": 73200
+      "learning_rate": 3.6893340624016176e-05,
+      "loss": 2.914,
+      "step": 79100
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.691776934581708e-05,
-      "loss": 2.8066,
-      "step": 73300
+      "learning_rate": 3.68767708902918e-05,
+      "loss": 2.9316,
+      "step": 79200
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.6899921827871395e-05,
-      "loss": 2.8117,
-      "step": 73400
+      "learning_rate": 3.686020115656741e-05,
+      "loss": 2.9347,
+      "step": 79300
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.6882074309925716e-05,
-      "loss": 2.8091,
-      "step": 73500
+      "learning_rate": 3.684363142284304e-05,
+      "loss": 2.9219,
+      "step": 79400
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.686422679198004e-05,
-      "loss": 2.8075,
-      "step": 73600
+      "learning_rate": 3.6827061689118655e-05,
+      "loss": 2.9354,
+      "step": 79500
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.684637927403436e-05,
-      "loss": 2.8055,
-      "step": 73700
+      "learning_rate": 3.681049195539428e-05,
+      "loss": 2.933,
+      "step": 79600
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.682853175608868e-05,
-      "loss": 2.8076,
-      "step": 73800
+      "learning_rate": 3.6793922221669905e-05,
+      "loss": 2.9381,
+      "step": 79700
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.6810684238143e-05,
-      "loss": 2.8159,
-      "step": 73900
+      "learning_rate": 3.677735248794552e-05,
+      "loss": 2.9332,
+      "step": 79800
     },
     {
       "epoch": 0.26,
-      "learning_rate": 3.679283672019732e-05,
-      "loss": 2.8066,
-      "step": 74000
+      "learning_rate": 3.676078275422114e-05,
+      "loss": 2.9256,
+      "step": 79900
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 3.677498920225165e-05,
-      "loss": 2.8005,
-      "step": 74100
+      "epoch": 0.27,
+      "learning_rate": 3.674421302049676e-05,
+      "loss": 2.9395,
+      "step": 80000
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 3.675714168430597e-05,
-      "loss": 2.809,
-      "step": 74200
+      "epoch": 0.27,
+      "learning_rate": 3.6727643286772384e-05,
+      "loss": 2.9218,
+      "step": 80100
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6739294166360286e-05,
-      "loss": 2.81,
-      "step": 74300
+      "learning_rate": 3.6711073553048006e-05,
+      "loss": 2.9237,
+      "step": 80200
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.672144664841461e-05,
-      "loss": 2.7959,
-      "step": 74400
+      "learning_rate": 3.669450381932363e-05,
+      "loss": 2.9395,
+      "step": 80300
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.670359913046893e-05,
-      "loss": 2.7977,
-      "step": 74500
+      "learning_rate": 3.667793408559925e-05,
+      "loss": 2.9189,
+      "step": 80400
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.668575161252325e-05,
-      "loss": 2.8063,
-      "step": 74600
+      "learning_rate": 3.666136435187486e-05,
+      "loss": 2.9255,
+      "step": 80500
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.666790409457757e-05,
-      "loss": 2.811,
-      "step": 74700
+      "learning_rate": 3.6644794618150485e-05,
+      "loss": 2.9297,
+      "step": 80600
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.665005657663189e-05,
-      "loss": 2.8095,
-      "step": 74800
+      "learning_rate": 3.662822488442611e-05,
+      "loss": 2.9378,
+      "step": 80700
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.663220905868621e-05,
-      "loss": 2.81,
-      "step": 74900
+      "learning_rate": 3.661165515070173e-05,
+      "loss": 2.9228,
+      "step": 80800
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6614361540740534e-05,
-      "loss": 2.8034,
-      "step": 75000
+      "learning_rate": 3.659508541697735e-05,
+      "loss": 2.9345,
+      "step": 80900
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6596514022794855e-05,
-      "loss": 2.8054,
-      "step": 75100
+      "learning_rate": 3.657851568325297e-05,
+      "loss": 2.936,
+      "step": 81000
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6578666504849177e-05,
-      "loss": 2.8026,
-      "step": 75200
+      "learning_rate": 3.656194594952859e-05,
+      "loss": 2.925,
+      "step": 81100
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.65608189869035e-05,
-      "loss": 2.8075,
-      "step": 75300
+      "learning_rate": 3.6545376215804213e-05,
+      "loss": 2.9329,
+      "step": 81200
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.654297146895781e-05,
-      "loss": 2.8087,
-      "step": 75400
+      "learning_rate": 3.6528806482079835e-05,
+      "loss": 2.9273,
+      "step": 81300
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6525123951012133e-05,
-      "loss": 2.7983,
-      "step": 75500
+      "learning_rate": 3.6512236748355456e-05,
+      "loss": 2.9247,
+      "step": 81400
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6507276433066455e-05,
-      "loss": 2.8036,
-      "step": 75600
+      "learning_rate": 3.649566701463107e-05,
+      "loss": 2.9335,
+      "step": 81500
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6489428915120776e-05,
-      "loss": 2.8184,
-      "step": 75700
+      "learning_rate": 3.64790972809067e-05,
+      "loss": 2.9322,
+      "step": 81600
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.64715813971751e-05,
-      "loss": 2.7996,
-      "step": 75800
+      "learning_rate": 3.646252754718232e-05,
+      "loss": 2.9228,
+      "step": 81700
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.645373387922942e-05,
-      "loss": 2.8028,
-      "step": 75900
+      "learning_rate": 3.6445957813457936e-05,
+      "loss": 2.9277,
+      "step": 81800
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.643588636128374e-05,
-      "loss": 2.8051,
-      "step": 76000
+      "learning_rate": 3.6429388079733564e-05,
+      "loss": 2.9139,
+      "step": 81900
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.641803884333806e-05,
-      "loss": 2.7947,
-      "step": 76100
+      "learning_rate": 3.641281834600918e-05,
+      "loss": 2.9377,
+      "step": 82000
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.640019132539238e-05,
-      "loss": 2.8027,
-      "step": 76200
+      "learning_rate": 3.63962486122848e-05,
+      "loss": 2.9294,
+      "step": 82100
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.63823438074467e-05,
-      "loss": 2.7995,
-      "step": 76300
+      "learning_rate": 3.637967887856043e-05,
+      "loss": 2.9265,
+      "step": 82200
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6364496289501024e-05,
-      "loss": 2.8027,
-      "step": 76400
+      "learning_rate": 3.636310914483604e-05,
+      "loss": 2.9282,
+      "step": 82300
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.634664877155534e-05,
-      "loss": 2.813,
-      "step": 76500
+      "learning_rate": 3.6346539411111664e-05,
+      "loss": 2.9292,
+      "step": 82400
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.632880125360966e-05,
-      "loss": 2.7952,
-      "step": 76600
+      "learning_rate": 3.6329969677387286e-05,
+      "loss": 2.9271,
+      "step": 82500
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.631095373566398e-05,
-      "loss": 2.8047,
-      "step": 76700
+      "learning_rate": 3.631339994366291e-05,
+      "loss": 2.9317,
+      "step": 82600
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.62931062177183e-05,
-      "loss": 2.8081,
-      "step": 76800
+      "learning_rate": 3.629683020993853e-05,
+      "loss": 2.9314,
+      "step": 82700
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6275258699772624e-05,
-      "loss": 2.8098,
-      "step": 76900
+      "learning_rate": 3.628026047621415e-05,
+      "loss": 2.9285,
+      "step": 82800
     },
     {
       "epoch": 0.27,
-      "learning_rate": 3.6257411181826945e-05,
-      "loss": 2.8081,
-      "step": 77000
+      "learning_rate": 3.626369074248977e-05,
+      "loss": 2.932,
+      "step": 82900
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.6239563663881266e-05,
-      "loss": 2.8029,
-      "step": 77100
+      "learning_rate": 3.6247121008765386e-05,
+      "loss": 2.925,
+      "step": 83000
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.622171614593559e-05,
-      "loss": 2.8009,
-      "step": 77200
+      "learning_rate": 3.6230551275041015e-05,
+      "loss": 2.9212,
+      "step": 83100
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.620386862798991e-05,
-      "loss": 2.8057,
-      "step": 77300
+      "learning_rate": 3.6213981541316636e-05,
+      "loss": 2.9178,
+      "step": 83200
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.618602111004423e-05,
-      "loss": 2.8088,
-      "step": 77400
+      "learning_rate": 3.619741180759225e-05,
+      "loss": 2.9304,
+      "step": 83300
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.616817359209855e-05,
-      "loss": 2.8007,
-      "step": 77500
+      "learning_rate": 3.618084207386788e-05,
+      "loss": 2.927,
+      "step": 83400
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.6150326074152865e-05,
-      "loss": 2.8024,
-      "step": 77600
+      "learning_rate": 3.6164272340143494e-05,
+      "loss": 2.9226,
+      "step": 83500
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.6132478556207187e-05,
-      "loss": 2.7984,
-      "step": 77700
+      "learning_rate": 3.6147702606419115e-05,
+      "loss": 2.9139,
+      "step": 83600
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.611463103826151e-05,
-      "loss": 2.811,
-      "step": 77800
+      "learning_rate": 3.613113287269474e-05,
+      "loss": 2.9279,
+      "step": 83700
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.609678352031583e-05,
-      "loss": 2.7933,
-      "step": 77900
+      "learning_rate": 3.611456313897036e-05,
+      "loss": 2.9177,
+      "step": 83800
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.607893600237015e-05,
-      "loss": 2.7904,
-      "step": 78000
+      "learning_rate": 3.609799340524598e-05,
+      "loss": 2.9319,
+      "step": 83900
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.606108848442447e-05,
-      "loss": 2.8006,
-      "step": 78100
+      "learning_rate": 3.60814236715216e-05,
+      "loss": 2.9398,
+      "step": 84000
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.604324096647879e-05,
-      "loss": 2.8043,
-      "step": 78200
+      "learning_rate": 3.606485393779722e-05,
+      "loss": 2.9292,
+      "step": 84100
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.6025393448533114e-05,
-      "loss": 2.7954,
-      "step": 78300
+      "learning_rate": 3.6048284204072844e-05,
+      "loss": 2.9214,
+      "step": 84200
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.6007545930587435e-05,
-      "loss": 2.7991,
-      "step": 78400
+      "learning_rate": 3.6031714470348465e-05,
+      "loss": 2.934,
+      "step": 84300
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5989698412641756e-05,
-      "loss": 2.8027,
-      "step": 78500
+      "learning_rate": 3.601514473662409e-05,
+      "loss": 2.9199,
+      "step": 84400
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.597185089469608e-05,
-      "loss": 2.8098,
-      "step": 78600
+      "learning_rate": 3.59985750028997e-05,
+      "loss": 2.9352,
+      "step": 84500
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.595400337675039e-05,
-      "loss": 2.8082,
-      "step": 78700
+      "learning_rate": 3.598200526917533e-05,
+      "loss": 2.9356,
+      "step": 84600
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.593615585880471e-05,
-      "loss": 2.7968,
-      "step": 78800
+      "learning_rate": 3.5965435535450945e-05,
+      "loss": 2.9341,
+      "step": 84700
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5918308340859034e-05,
-      "loss": 2.8025,
-      "step": 78900
+      "learning_rate": 3.5948865801726566e-05,
+      "loss": 2.9133,
+      "step": 84800
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5900460822913356e-05,
-      "loss": 2.7976,
-      "step": 79000
+      "learning_rate": 3.5932296068002194e-05,
+      "loss": 2.922,
+      "step": 84900
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.588261330496768e-05,
-      "loss": 2.8049,
-      "step": 79100
+      "learning_rate": 3.591572633427781e-05,
+      "loss": 2.9391,
+      "step": 85000
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5864765787022005e-05,
-      "loss": 2.8057,
-      "step": 79200
+      "learning_rate": 3.589915660055343e-05,
+      "loss": 2.92,
+      "step": 85100
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5846918269076326e-05,
-      "loss": 2.8104,
-      "step": 79300
+      "learning_rate": 3.588258686682905e-05,
+      "loss": 2.9264,
+      "step": 85200
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.582907075113065e-05,
-      "loss": 2.796,
-      "step": 79400
+      "learning_rate": 3.586601713310467e-05,
+      "loss": 2.9205,
+      "step": 85300
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.581122323318497e-05,
-      "loss": 2.8066,
-      "step": 79500
+      "learning_rate": 3.5849447399380295e-05,
+      "loss": 2.9131,
+      "step": 85400
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.579337571523928e-05,
-      "loss": 2.8091,
-      "step": 79600
+      "learning_rate": 3.5832877665655916e-05,
+      "loss": 2.925,
+      "step": 85500
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5775528197293604e-05,
-      "loss": 2.8055,
-      "step": 79700
+      "learning_rate": 3.581630793193154e-05,
+      "loss": 2.914,
+      "step": 85600
     },
     {
       "epoch": 0.28,
-      "learning_rate": 3.5757680679347925e-05,
-      "loss": 2.7998,
-      "step": 79800
+      "learning_rate": 3.579973819820715e-05,
+      "loss": 2.9307,
+      "step": 85700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5783168464482774e-05,
+      "loss": 2.9285,
+      "step": 85800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.57665987307584e-05,
+      "loss": 2.9225,
+      "step": 85900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.575002899703402e-05,
+      "loss": 2.9222,
+      "step": 86000
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5739833161402247e-05,
-      "loss": 2.7972,
-      "step": 79900
+      "learning_rate": 3.573345926330964e-05,
+      "loss": 2.9263,
+      "step": 86100
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.572198564345657e-05,
-      "loss": 2.8224,
-      "step": 80000
+      "learning_rate": 3.571688952958526e-05,
+      "loss": 2.9241,
+      "step": 86200
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.570413812551089e-05,
-      "loss": 2.7974,
-      "step": 80100
+      "learning_rate": 3.570031979586088e-05,
+      "loss": 2.9265,
+      "step": 86300
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.568629060756521e-05,
-      "loss": 2.803,
-      "step": 80200
+      "learning_rate": 3.56837500621365e-05,
+      "loss": 2.9218,
+      "step": 86400
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.566844308961953e-05,
-      "loss": 2.8,
-      "step": 80300
+      "learning_rate": 3.5667180328412124e-05,
+      "loss": 2.9173,
+      "step": 86500
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.565059557167385e-05,
-      "loss": 2.8029,
-      "step": 80400
+      "learning_rate": 3.5650610594687746e-05,
+      "loss": 2.9314,
+      "step": 86600
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5632748053728174e-05,
-      "loss": 2.7944,
-      "step": 80500
+      "learning_rate": 3.563404086096336e-05,
+      "loss": 2.9328,
+      "step": 86700
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.561490053578249e-05,
-      "loss": 2.7996,
-      "step": 80600
+      "learning_rate": 3.561747112723899e-05,
+      "loss": 2.9084,
+      "step": 86800
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.559705301783681e-05,
-      "loss": 2.8045,
-      "step": 80700
+      "learning_rate": 3.560090139351461e-05,
+      "loss": 2.9424,
+      "step": 86900
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.557920549989113e-05,
-      "loss": 2.8062,
-      "step": 80800
+      "learning_rate": 3.5584331659790225e-05,
+      "loss": 2.9225,
+      "step": 87000
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.556135798194545e-05,
-      "loss": 2.7918,
-      "step": 80900
+      "learning_rate": 3.556776192606585e-05,
+      "loss": 2.9213,
+      "step": 87100
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.554351046399977e-05,
-      "loss": 2.7998,
-      "step": 81000
+      "learning_rate": 3.555119219234147e-05,
+      "loss": 2.9152,
+      "step": 87200
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5525662946054094e-05,
-      "loss": 2.8071,
-      "step": 81100
+      "learning_rate": 3.553462245861709e-05,
+      "loss": 2.9272,
+      "step": 87300
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5507815428108416e-05,
-      "loss": 2.7921,
-      "step": 81200
+      "learning_rate": 3.551805272489272e-05,
+      "loss": 2.9436,
+      "step": 87400
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.548996791016274e-05,
-      "loss": 2.7935,
-      "step": 81300
+      "learning_rate": 3.550148299116833e-05,
+      "loss": 2.9169,
+      "step": 87500
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.547212039221706e-05,
-      "loss": 2.7901,
-      "step": 81400
+      "learning_rate": 3.5484913257443954e-05,
+      "loss": 2.9185,
+      "step": 87600
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.545427287427138e-05,
-      "loss": 2.7961,
-      "step": 81500
+      "learning_rate": 3.5468343523719575e-05,
+      "loss": 2.9304,
+      "step": 87700
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.54364253563257e-05,
-      "loss": 2.8023,
-      "step": 81600
+      "learning_rate": 3.5451773789995196e-05,
+      "loss": 2.9285,
+      "step": 87800
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5418577838380015e-05,
-      "loss": 2.7976,
-      "step": 81700
+      "learning_rate": 3.543520405627082e-05,
+      "loss": 2.9331,
+      "step": 87900
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5400730320434336e-05,
-      "loss": 2.8003,
-      "step": 81800
+      "learning_rate": 3.541863432254644e-05,
+      "loss": 2.918,
+      "step": 88000
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.538288280248866e-05,
-      "loss": 2.7947,
-      "step": 81900
+      "learning_rate": 3.540206458882206e-05,
+      "loss": 2.9198,
+      "step": 88100
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.536503528454298e-05,
-      "loss": 2.7876,
-      "step": 82000
+      "learning_rate": 3.5385494855097676e-05,
+      "loss": 2.9256,
+      "step": 88200
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.53471877665973e-05,
-      "loss": 2.8015,
-      "step": 82100
+      "learning_rate": 3.5368925121373304e-05,
+      "loss": 2.9182,
+      "step": 88300
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.532934024865162e-05,
-      "loss": 2.8107,
-      "step": 82200
+      "learning_rate": 3.5352355387648925e-05,
+      "loss": 2.9075,
+      "step": 88400
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.531149273070594e-05,
-      "loss": 2.7999,
-      "step": 82300
+      "learning_rate": 3.533578565392454e-05,
+      "loss": 2.9194,
+      "step": 88500
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.529364521276026e-05,
-      "loss": 2.795,
-      "step": 82400
+      "learning_rate": 3.531921592020017e-05,
+      "loss": 2.9365,
+      "step": 88600
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5275797694814585e-05,
-      "loss": 2.7994,
-      "step": 82500
+      "learning_rate": 3.530264618647578e-05,
+      "loss": 2.9166,
+      "step": 88700
     },
     {
       "epoch": 0.29,
-      "learning_rate": 3.5257950176868906e-05,
-      "loss": 2.8099,
-      "step": 82600
+      "learning_rate": 3.5286076452751404e-05,
+      "loss": 2.9235,
+      "step": 88800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5269506719027026e-05,
+      "loss": 2.9105,
+      "step": 88900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.525293698530265e-05,
+      "loss": 2.928,
+      "step": 89000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.523636725157827e-05,
+      "loss": 2.9173,
+      "step": 89100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.521979751785389e-05,
+      "loss": 2.9134,
+      "step": 89200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.520322778412951e-05,
+      "loss": 2.9298,
+      "step": 89300
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.524010265892323e-05,
-      "loss": 2.8062,
-      "step": 82700
+      "learning_rate": 3.518665805040513e-05,
+      "loss": 2.9121,
+      "step": 89400
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.522225514097754e-05,
-      "loss": 2.795,
-      "step": 82800
+      "learning_rate": 3.5170088316680755e-05,
+      "loss": 2.9251,
+      "step": 89500
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.520440762303186e-05,
-      "loss": 2.8096,
-      "step": 82900
+      "learning_rate": 3.5153518582956376e-05,
+      "loss": 2.913,
+      "step": 89600
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.5186560105086184e-05,
-      "loss": 2.7915,
-      "step": 83000
+      "learning_rate": 3.513694884923199e-05,
+      "loss": 2.9197,
+      "step": 89700
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.5168712587140505e-05,
-      "loss": 2.7943,
-      "step": 83100
+      "learning_rate": 3.512037911550762e-05,
+      "loss": 2.9184,
+      "step": 89800
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.5150865069194826e-05,
-      "loss": 2.7945,
-      "step": 83200
+      "learning_rate": 3.5103809381783234e-05,
+      "loss": 2.9378,
+      "step": 89900
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.513301755124915e-05,
-      "loss": 2.796,
-      "step": 83300
+      "learning_rate": 3.5087239648058855e-05,
+      "loss": 2.9296,
+      "step": 90000
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.511517003330347e-05,
-      "loss": 2.8073,
-      "step": 83400
+      "learning_rate": 3.5070669914334483e-05,
+      "loss": 2.9214,
+      "step": 90100
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.509732251535779e-05,
-      "loss": 2.8,
-      "step": 83500
+      "learning_rate": 3.50541001806101e-05,
+      "loss": 2.9224,
+      "step": 90200
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.507947499741211e-05,
-      "loss": 2.7775,
-      "step": 83600
+      "learning_rate": 3.503753044688572e-05,
+      "loss": 2.9306,
+      "step": 90300
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.506162747946643e-05,
-      "loss": 2.8005,
-      "step": 83700
+      "learning_rate": 3.502096071316134e-05,
+      "loss": 2.9204,
+      "step": 90400
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.5043779961520754e-05,
-      "loss": 2.7855,
-      "step": 83800
+      "learning_rate": 3.500439097943696e-05,
+      "loss": 2.9102,
+      "step": 90500
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.502593244357507e-05,
-      "loss": 2.7942,
-      "step": 83900
+      "learning_rate": 3.4987821245712584e-05,
+      "loss": 2.9058,
+      "step": 90600
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.500808492562939e-05,
-      "loss": 2.7927,
-      "step": 84000
+      "learning_rate": 3.4971251511988205e-05,
+      "loss": 2.9233,
+      "step": 90700
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.499023740768371e-05,
-      "loss": 2.7909,
-      "step": 84100
+      "learning_rate": 3.495468177826383e-05,
+      "loss": 2.9299,
+      "step": 90800
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.497238988973803e-05,
-      "loss": 2.7891,
-      "step": 84200
+      "learning_rate": 3.493811204453944e-05,
+      "loss": 2.9164,
+      "step": 90900
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.495454237179236e-05,
-      "loss": 2.7811,
-      "step": 84300
+      "learning_rate": 3.492154231081506e-05,
+      "loss": 2.9335,
+      "step": 91000
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.493669485384668e-05,
-      "loss": 2.7972,
-      "step": 84400
+      "learning_rate": 3.490497257709069e-05,
+      "loss": 2.9161,
+      "step": 91100
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.4918847335901e-05,
-      "loss": 2.7969,
-      "step": 84500
+      "learning_rate": 3.4888402843366306e-05,
+      "loss": 2.9219,
+      "step": 91200
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.490099981795532e-05,
-      "loss": 2.786,
-      "step": 84600
+      "learning_rate": 3.487183310964193e-05,
+      "loss": 2.9129,
+      "step": 91300
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.4883152300009645e-05,
-      "loss": 2.7791,
-      "step": 84700
+      "learning_rate": 3.485526337591755e-05,
+      "loss": 2.9266,
+      "step": 91400
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.486530478206396e-05,
-      "loss": 2.7897,
-      "step": 84800
+      "learning_rate": 3.483869364219317e-05,
+      "loss": 2.9213,
+      "step": 91500
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.484745726411828e-05,
-      "loss": 2.7829,
-      "step": 84900
+      "learning_rate": 3.482212390846879e-05,
+      "loss": 2.9092,
+      "step": 91600
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.48296097461726e-05,
-      "loss": 2.7877,
-      "step": 85000
+      "learning_rate": 3.480555417474441e-05,
+      "loss": 2.936,
+      "step": 91700
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.481176222822692e-05,
-      "loss": 2.7965,
-      "step": 85100
+      "learning_rate": 3.4788984441020035e-05,
+      "loss": 2.916,
+      "step": 91800
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.4793914710281244e-05,
-      "loss": 2.7986,
-      "step": 85200
+      "learning_rate": 3.4772414707295656e-05,
+      "loss": 2.8989,
+      "step": 91900
     },
     {
       "epoch": 0.3,
-      "learning_rate": 3.4776067192335565e-05,
-      "loss": 2.7859,
-      "step": 85300
+      "learning_rate": 3.475584497357128e-05,
+      "loss": 2.926,
+      "step": 92000
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 3.4758219674389886e-05,
-      "loss": 2.7989,
-      "step": 85400
+      "epoch": 0.31,
+      "learning_rate": 3.47392752398469e-05,
+      "loss": 2.9135,
+      "step": 92100
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.474037215644421e-05,
-      "loss": 2.7942,
-      "step": 85500
+      "learning_rate": 3.4722705506122514e-05,
+      "loss": 2.9041,
+      "step": 92200
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.472252463849853e-05,
-      "loss": 2.793,
-      "step": 85600
+      "learning_rate": 3.470613577239814e-05,
+      "loss": 2.9128,
+      "step": 92300
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.470467712055285e-05,
-      "loss": 2.7894,
-      "step": 85700
+      "learning_rate": 3.468956603867376e-05,
+      "loss": 2.9234,
+      "step": 92400
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.468682960260717e-05,
-      "loss": 2.7974,
-      "step": 85800
+      "learning_rate": 3.467299630494938e-05,
+      "loss": 2.9208,
+      "step": 92500
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4668982084661486e-05,
-      "loss": 2.7912,
-      "step": 85900
+      "learning_rate": 3.4656426571225007e-05,
+      "loss": 2.9228,
+      "step": 92600
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.465113456671581e-05,
-      "loss": 2.8016,
-      "step": 86000
+      "learning_rate": 3.463985683750062e-05,
+      "loss": 2.9189,
+      "step": 92700
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.463328704877013e-05,
-      "loss": 2.7912,
-      "step": 86100
+      "learning_rate": 3.462328710377624e-05,
+      "loss": 2.9055,
+      "step": 92800
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.461543953082445e-05,
-      "loss": 2.7997,
-      "step": 86200
+      "learning_rate": 3.4606717370051864e-05,
+      "loss": 2.9313,
+      "step": 92900
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.459759201287877e-05,
-      "loss": 2.7946,
-      "step": 86300
+      "learning_rate": 3.4590147636327486e-05,
+      "loss": 2.912,
+      "step": 93000
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.457974449493309e-05,
-      "loss": 2.7986,
-      "step": 86400
+      "learning_rate": 3.457357790260311e-05,
+      "loss": 2.9258,
+      "step": 93100
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.456189697698741e-05,
-      "loss": 2.7806,
-      "step": 86500
+      "learning_rate": 3.455700816887873e-05,
+      "loss": 2.9068,
+      "step": 93200
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4544049459041734e-05,
-      "loss": 2.8009,
-      "step": 86600
+      "learning_rate": 3.454043843515435e-05,
+      "loss": 2.9183,
+      "step": 93300
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4526201941096055e-05,
-      "loss": 2.7881,
-      "step": 86700
+      "learning_rate": 3.4523868701429965e-05,
+      "loss": 2.9199,
+      "step": 93400
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4508354423150376e-05,
-      "loss": 2.7896,
-      "step": 86800
+      "learning_rate": 3.450729896770559e-05,
+      "loss": 2.9143,
+      "step": 93500
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.44905069052047e-05,
-      "loss": 2.7916,
-      "step": 86900
+      "learning_rate": 3.4490729233981214e-05,
+      "loss": 2.9342,
+      "step": 93600
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.447265938725901e-05,
-      "loss": 2.7909,
-      "step": 87000
+      "learning_rate": 3.447415950025683e-05,
+      "loss": 2.9164,
+      "step": 93700
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.445481186931333e-05,
-      "loss": 2.7981,
-      "step": 87100
+      "learning_rate": 3.445758976653246e-05,
+      "loss": 2.9072,
+      "step": 93800
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4436964351367655e-05,
-      "loss": 2.7981,
-      "step": 87200
+      "learning_rate": 3.444102003280807e-05,
+      "loss": 2.9188,
+      "step": 93900
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4419116833421976e-05,
-      "loss": 2.7944,
-      "step": 87300
+      "learning_rate": 3.4424450299083694e-05,
+      "loss": 2.9236,
+      "step": 94000
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.44012693154763e-05,
-      "loss": 2.7774,
-      "step": 87400
+      "learning_rate": 3.4407880565359315e-05,
+      "loss": 2.9229,
+      "step": 94100
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.438342179753062e-05,
-      "loss": 2.7957,
-      "step": 87500
+      "learning_rate": 3.4391310831634936e-05,
+      "loss": 2.9135,
+      "step": 94200
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.436557427958494e-05,
-      "loss": 2.7897,
-      "step": 87600
+      "learning_rate": 3.437474109791056e-05,
+      "loss": 2.9248,
+      "step": 94300
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.434772676163926e-05,
-      "loss": 2.7931,
-      "step": 87700
+      "learning_rate": 3.435817136418618e-05,
+      "loss": 2.917,
+      "step": 94400
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.432987924369358e-05,
-      "loss": 2.8006,
-      "step": 87800
+      "learning_rate": 3.43416016304618e-05,
+      "loss": 2.899,
+      "step": 94500
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.43120317257479e-05,
-      "loss": 2.7896,
-      "step": 87900
+      "learning_rate": 3.432503189673742e-05,
+      "loss": 2.9203,
+      "step": 94600
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.4294184207802224e-05,
-      "loss": 2.7926,
-      "step": 88000
+      "learning_rate": 3.4308462163013044e-05,
+      "loss": 2.9191,
+      "step": 94700
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.427633668985654e-05,
-      "loss": 2.7889,
-      "step": 88100
+      "learning_rate": 3.4291892429288665e-05,
+      "loss": 2.918,
+      "step": 94800
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.425848917191086e-05,
-      "loss": 2.7929,
-      "step": 88200
+      "learning_rate": 3.427532269556428e-05,
+      "loss": 2.9168,
+      "step": 94900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.425875296183991e-05,
+      "loss": 2.9085,
+      "step": 95000
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.424064165396518e-05,
-      "loss": 2.8003,
-      "step": 88300
+      "learning_rate": 3.424218322811553e-05,
+      "loss": 2.9222,
+      "step": 95100
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.42227941360195e-05,
-      "loss": 2.7875,
-      "step": 88400
+      "learning_rate": 3.4225613494391144e-05,
+      "loss": 2.9237,
+      "step": 95200
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4204946618073824e-05,
-      "loss": 2.7897,
-      "step": 88500
+      "learning_rate": 3.420904376066677e-05,
+      "loss": 2.925,
+      "step": 95300
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4187099100128145e-05,
-      "loss": 2.7913,
-      "step": 88600
+      "learning_rate": 3.419247402694239e-05,
+      "loss": 2.9144,
+      "step": 95400
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4169251582182466e-05,
-      "loss": 2.7967,
-      "step": 88700
+      "learning_rate": 3.417590429321801e-05,
+      "loss": 2.9174,
+      "step": 95500
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.415140406423679e-05,
-      "loss": 2.7979,
-      "step": 88800
+      "learning_rate": 3.415933455949363e-05,
+      "loss": 2.9127,
+      "step": 95600
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.413355654629111e-05,
-      "loss": 2.7955,
-      "step": 88900
+      "learning_rate": 3.414276482576925e-05,
+      "loss": 2.9136,
+      "step": 95700
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.411570902834543e-05,
-      "loss": 2.7939,
-      "step": 89000
+      "learning_rate": 3.412619509204487e-05,
+      "loss": 2.9234,
+      "step": 95800
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.409786151039975e-05,
-      "loss": 2.7744,
-      "step": 89100
+      "learning_rate": 3.4109625358320495e-05,
+      "loss": 2.9275,
+      "step": 95900
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4080013992454065e-05,
-      "loss": 2.7845,
-      "step": 89200
+      "learning_rate": 3.4093055624596116e-05,
+      "loss": 2.9146,
+      "step": 96000
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4062166474508387e-05,
-      "loss": 2.7793,
-      "step": 89300
+      "learning_rate": 3.407648589087174e-05,
+      "loss": 2.9218,
+      "step": 96100
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4044318956562715e-05,
-      "loss": 2.7996,
-      "step": 89400
+      "learning_rate": 3.405991615714735e-05,
+      "loss": 2.9143,
+      "step": 96200
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.4026471438617036e-05,
-      "loss": 2.7876,
-      "step": 89500
+      "learning_rate": 3.404334642342298e-05,
+      "loss": 2.9046,
+      "step": 96300
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.400862392067136e-05,
-      "loss": 2.7879,
-      "step": 89600
+      "learning_rate": 3.4026776689698595e-05,
+      "loss": 2.9288,
+      "step": 96400
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.399077640272568e-05,
-      "loss": 2.7849,
-      "step": 89700
+      "learning_rate": 3.401020695597422e-05,
+      "loss": 2.9323,
+      "step": 96500
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.397292888478e-05,
-      "loss": 2.7854,
-      "step": 89800
+      "learning_rate": 3.399363722224984e-05,
+      "loss": 2.9317,
+      "step": 96600
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.395508136683432e-05,
-      "loss": 2.7948,
-      "step": 89900
+      "learning_rate": 3.397706748852546e-05,
+      "loss": 2.9171,
+      "step": 96700
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.393723384888864e-05,
-      "loss": 2.7837,
-      "step": 90000
+      "learning_rate": 3.396049775480108e-05,
+      "loss": 2.9227,
+      "step": 96800
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.3919386330942956e-05,
-      "loss": 2.7937,
-      "step": 90100
+      "learning_rate": 3.39439280210767e-05,
+      "loss": 2.9137,
+      "step": 96900
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.390153881299728e-05,
-      "loss": 2.787,
-      "step": 90200
+      "learning_rate": 3.3927358287352324e-05,
+      "loss": 2.9193,
+      "step": 97000
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.38836912950516e-05,
-      "loss": 2.7907,
-      "step": 90300
+      "learning_rate": 3.3910788553627945e-05,
+      "loss": 2.9102,
+      "step": 97100
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.386584377710592e-05,
-      "loss": 2.7884,
-      "step": 90400
+      "learning_rate": 3.389421881990357e-05,
+      "loss": 2.9128,
+      "step": 97200
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.384799625916024e-05,
-      "loss": 2.7879,
-      "step": 90500
+      "learning_rate": 3.387764908617919e-05,
+      "loss": 2.9223,
+      "step": 97300
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.383014874121456e-05,
-      "loss": 2.8006,
-      "step": 90600
+      "learning_rate": 3.38610793524548e-05,
+      "loss": 2.9219,
+      "step": 97400
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.3812301223268884e-05,
-      "loss": 2.8,
-      "step": 90700
+      "learning_rate": 3.384450961873043e-05,
+      "loss": 2.936,
+      "step": 97500
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.3794453705323205e-05,
-      "loss": 2.7887,
-      "step": 90800
+      "learning_rate": 3.3827939885006046e-05,
+      "loss": 2.9141,
+      "step": 97600
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.3776606187377526e-05,
-      "loss": 2.7826,
-      "step": 90900
+      "learning_rate": 3.381137015128167e-05,
+      "loss": 2.9201,
+      "step": 97700
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.375875866943185e-05,
-      "loss": 2.7828,
-      "step": 91000
+      "learning_rate": 3.3794800417557296e-05,
+      "loss": 2.9242,
+      "step": 97800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.377823068383291e-05,
+      "loss": 2.9187,
+      "step": 97900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.376166095010853e-05,
+      "loss": 2.922,
+      "step": 98000
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.374091115148617e-05,
-      "loss": 2.7941,
-      "step": 91100
+      "learning_rate": 3.374509121638415e-05,
+      "loss": 2.9184,
+      "step": 98100
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.372306363354048e-05,
-      "loss": 2.7942,
-      "step": 91200
+      "learning_rate": 3.3728521482659775e-05,
+      "loss": 2.9224,
+      "step": 98200
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3705216115594804e-05,
-      "loss": 2.7859,
-      "step": 91300
+      "learning_rate": 3.3711951748935396e-05,
+      "loss": 2.9107,
+      "step": 98300
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3687368597649125e-05,
-      "loss": 2.7765,
-      "step": 91400
+      "learning_rate": 3.369538201521102e-05,
+      "loss": 2.9157,
+      "step": 98400
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3669521079703446e-05,
-      "loss": 2.7843,
-      "step": 91500
+      "learning_rate": 3.367881228148664e-05,
+      "loss": 2.9226,
+      "step": 98500
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.365167356175777e-05,
-      "loss": 2.7845,
-      "step": 91600
+      "learning_rate": 3.3662242547762254e-05,
+      "loss": 2.9262,
+      "step": 98600
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.363382604381209e-05,
-      "loss": 2.7959,
-      "step": 91700
+      "learning_rate": 3.364567281403788e-05,
+      "loss": 2.9038,
+      "step": 98700
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.361597852586641e-05,
-      "loss": 2.7833,
-      "step": 91800
+      "learning_rate": 3.3629103080313504e-05,
+      "loss": 2.9034,
+      "step": 98800
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.359813100792073e-05,
-      "loss": 2.7813,
-      "step": 91900
+      "learning_rate": 3.361253334658912e-05,
+      "loss": 2.9092,
+      "step": 98900
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.358028348997505e-05,
-      "loss": 2.7833,
-      "step": 92000
+      "learning_rate": 3.3595963612864747e-05,
+      "loss": 2.9153,
+      "step": 99000
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3562435972029374e-05,
-      "loss": 2.7833,
-      "step": 92100
+      "learning_rate": 3.357939387914036e-05,
+      "loss": 2.9123,
+      "step": 99100
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3544588454083695e-05,
-      "loss": 2.7922,
-      "step": 92200
+      "learning_rate": 3.356282414541598e-05,
+      "loss": 2.9199,
+      "step": 99200
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.352674093613801e-05,
-      "loss": 2.7835,
-      "step": 92300
+      "learning_rate": 3.354625441169161e-05,
+      "loss": 2.9014,
+      "step": 99300
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.350889341819233e-05,
-      "loss": 2.7863,
-      "step": 92400
+      "learning_rate": 3.3529684677967226e-05,
+      "loss": 2.9246,
+      "step": 99400
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.349104590024665e-05,
-      "loss": 2.7915,
-      "step": 92500
+      "learning_rate": 3.351311494424285e-05,
+      "loss": 2.9027,
+      "step": 99500
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.347319838230097e-05,
-      "loss": 2.7777,
-      "step": 92600
+      "learning_rate": 3.349654521051847e-05,
+      "loss": 2.9043,
+      "step": 99600
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3455350864355294e-05,
-      "loss": 2.7776,
-      "step": 92700
+      "learning_rate": 3.347997547679409e-05,
+      "loss": 2.9117,
+      "step": 99700
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3437503346409615e-05,
-      "loss": 2.7845,
-      "step": 92800
+      "learning_rate": 3.346340574306971e-05,
+      "loss": 2.9183,
+      "step": 99800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.344683600934533e-05,
+      "loss": 2.9076,
+      "step": 99900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3430266275620954e-05,
+      "loss": 2.9035,
+      "step": 100000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.341369654189657e-05,
+      "loss": 2.9102,
+      "step": 100100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33971268081722e-05,
+      "loss": 2.9122,
+      "step": 100200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.338055707444782e-05,
+      "loss": 2.9207,
+      "step": 100300
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.341965582846394e-05,
-      "loss": 2.7885,
-      "step": 92900
+      "learning_rate": 3.3363987340723434e-05,
+      "loss": 2.9203,
+      "step": 100400
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.340180831051826e-05,
-      "loss": 2.7868,
-      "step": 93000
+      "learning_rate": 3.334741760699906e-05,
+      "loss": 2.9207,
+      "step": 100500
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.338396079257258e-05,
-      "loss": 2.7877,
-      "step": 93100
+      "learning_rate": 3.3330847873274676e-05,
+      "loss": 2.9194,
+      "step": 100600
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.33661132746269e-05,
-      "loss": 2.7766,
-      "step": 93200
+      "learning_rate": 3.33142781395503e-05,
+      "loss": 2.9128,
+      "step": 100700
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.334826575668122e-05,
-      "loss": 2.777,
-      "step": 93300
+      "learning_rate": 3.329770840582592e-05,
+      "loss": 2.9133,
+      "step": 100800
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.3330418238735536e-05,
-      "loss": 2.7879,
-      "step": 93400
+      "learning_rate": 3.328113867210154e-05,
+      "loss": 2.9204,
+      "step": 100900
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.331257072078986e-05,
-      "loss": 2.7862,
-      "step": 93500
+      "learning_rate": 3.326456893837716e-05,
+      "loss": 2.9144,
+      "step": 101000
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 3.329472320284418e-05,
-      "loss": 2.7689,
-      "step": 93600
+      "epoch": 0.34,
+      "learning_rate": 3.3247999204652784e-05,
+      "loss": 2.9161,
+      "step": 101100
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 3.32768756848985e-05,
-      "loss": 2.7883,
-      "step": 93700
+      "epoch": 0.34,
+      "learning_rate": 3.3231429470928405e-05,
+      "loss": 2.9105,
+      "step": 101200
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 3.325902816695282e-05,
-      "loss": 2.7862,
-      "step": 93800
+      "epoch": 0.34,
+      "learning_rate": 3.321485973720403e-05,
+      "loss": 2.901,
+      "step": 101300
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.324118064900714e-05,
-      "loss": 2.789,
-      "step": 93900
+      "learning_rate": 3.319829000347964e-05,
+      "loss": 2.9184,
+      "step": 101400
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.322333313106146e-05,
-      "loss": 2.7981,
-      "step": 94000
+      "learning_rate": 3.318172026975527e-05,
+      "loss": 2.9139,
+      "step": 101500
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.3205485613115785e-05,
-      "loss": 2.78,
-      "step": 94100
+      "learning_rate": 3.3165150536030884e-05,
+      "loss": 2.9045,
+      "step": 101600
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.3187638095170106e-05,
-      "loss": 2.7938,
-      "step": 94200
+      "learning_rate": 3.3148580802306506e-05,
+      "loss": 2.9174,
+      "step": 101700
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.316979057722443e-05,
-      "loss": 2.7848,
-      "step": 94300
+      "learning_rate": 3.313201106858213e-05,
+      "loss": 2.9177,
+      "step": 101800
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.315194305927875e-05,
-      "loss": 2.7788,
-      "step": 94400
+      "learning_rate": 3.311544133485775e-05,
+      "loss": 2.9235,
+      "step": 101900
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.313409554133307e-05,
-      "loss": 2.7731,
-      "step": 94500
+      "learning_rate": 3.309887160113337e-05,
+      "loss": 2.9167,
+      "step": 102000
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.311624802338739e-05,
-      "loss": 2.7879,
-      "step": 94600
+      "learning_rate": 3.308230186740899e-05,
+      "loss": 2.9083,
+      "step": 102100
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.309840050544171e-05,
-      "loss": 2.7804,
-      "step": 94700
+      "learning_rate": 3.306573213368461e-05,
+      "loss": 2.9108,
+      "step": 102200
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.308055298749603e-05,
-      "loss": 2.7745,
-      "step": 94800
+      "learning_rate": 3.3049162399960235e-05,
+      "loss": 2.911,
+      "step": 102300
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.3062705469550354e-05,
-      "loss": 2.7767,
-      "step": 94900
+      "learning_rate": 3.3032592666235856e-05,
+      "loss": 2.9093,
+      "step": 102400
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.3044857951604675e-05,
-      "loss": 2.7902,
-      "step": 95000
+      "learning_rate": 3.301602293251148e-05,
+      "loss": 2.9117,
+      "step": 102500
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.3027010433659e-05,
-      "loss": 2.7728,
-      "step": 95100
+      "learning_rate": 3.299945319878709e-05,
+      "loss": 2.9143,
+      "step": 102600
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.300916291571332e-05,
-      "loss": 2.7857,
-      "step": 95200
+      "learning_rate": 3.298288346506272e-05,
+      "loss": 2.9241,
+      "step": 102700
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.299131539776764e-05,
-      "loss": 2.7792,
-      "step": 95300
+      "learning_rate": 3.2966313731338335e-05,
+      "loss": 2.927,
+      "step": 102800
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.2973467879821954e-05,
-      "loss": 2.7858,
-      "step": 95400
+      "learning_rate": 3.294974399761396e-05,
+      "loss": 2.917,
+      "step": 102900
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.2955620361876275e-05,
-      "loss": 2.7917,
-      "step": 95500
+      "learning_rate": 3.2933174263889585e-05,
+      "loss": 2.9072,
+      "step": 103000
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.2937772843930596e-05,
-      "loss": 2.7761,
-      "step": 95600
+      "learning_rate": 3.29166045301652e-05,
+      "loss": 2.9172,
+      "step": 103100
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.291992532598492e-05,
-      "loss": 2.7859,
-      "step": 95700
+      "learning_rate": 3.290003479644082e-05,
+      "loss": 2.9016,
+      "step": 103200
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.290207780803924e-05,
-      "loss": 2.7807,
-      "step": 95800
+      "learning_rate": 3.288346506271644e-05,
+      "loss": 2.9202,
+      "step": 103300
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.288423029009356e-05,
-      "loss": 2.7925,
-      "step": 95900
+      "learning_rate": 3.2866895328992064e-05,
+      "loss": 2.8986,
+      "step": 103400
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.286638277214788e-05,
-      "loss": 2.7724,
-      "step": 96000
+      "learning_rate": 3.2850325595267685e-05,
+      "loss": 2.8933,
+      "step": 103500
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.28485352542022e-05,
-      "loss": 2.7807,
-      "step": 96100
+      "learning_rate": 3.283375586154331e-05,
+      "loss": 2.9054,
+      "step": 103600
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.283068773625652e-05,
-      "loss": 2.7829,
-      "step": 96200
+      "learning_rate": 3.281718612781893e-05,
+      "loss": 2.9117,
+      "step": 103700
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.2812840218310844e-05,
-      "loss": 2.7864,
-      "step": 96300
+      "learning_rate": 3.280061639409455e-05,
+      "loss": 2.9208,
+      "step": 103800
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.2794992700365166e-05,
-      "loss": 2.7896,
-      "step": 96400
+      "learning_rate": 3.278404666037017e-05,
+      "loss": 2.9165,
+      "step": 103900
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.277714518241948e-05,
-      "loss": 2.7769,
-      "step": 96500
+      "learning_rate": 3.276747692664579e-05,
+      "loss": 2.91,
+      "step": 104000
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.27592976644738e-05,
-      "loss": 2.7887,
-      "step": 96600
+      "learning_rate": 3.275090719292141e-05,
+      "loss": 2.9031,
+      "step": 104100
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.274145014652812e-05,
-      "loss": 2.7704,
-      "step": 96700
+      "learning_rate": 3.2734337459197036e-05,
+      "loss": 2.9094,
+      "step": 104200
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2723602628582444e-05,
-      "loss": 2.784,
-      "step": 96800
+      "learning_rate": 3.271776772547265e-05,
+      "loss": 2.9097,
+      "step": 104300
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2705755110636765e-05,
-      "loss": 2.7876,
-      "step": 96900
+      "learning_rate": 3.270119799174827e-05,
+      "loss": 2.9073,
+      "step": 104400
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2687907592691086e-05,
-      "loss": 2.7886,
-      "step": 97000
+      "learning_rate": 3.26846282580239e-05,
+      "loss": 2.9044,
+      "step": 104500
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.267006007474541e-05,
-      "loss": 2.7802,
-      "step": 97100
+      "learning_rate": 3.2668058524299515e-05,
+      "loss": 2.9141,
+      "step": 104600
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.265221255679973e-05,
-      "loss": 2.7813,
-      "step": 97200
+      "learning_rate": 3.2651488790575136e-05,
+      "loss": 2.9023,
+      "step": 104700
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.263436503885405e-05,
-      "loss": 2.7827,
-      "step": 97300
+      "learning_rate": 3.263491905685076e-05,
+      "loss": 2.9061,
+      "step": 104800
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.261651752090837e-05,
-      "loss": 2.7824,
-      "step": 97400
+      "learning_rate": 3.261834932312638e-05,
+      "loss": 2.9076,
+      "step": 104900
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.259867000296269e-05,
-      "loss": 2.7709,
-      "step": 97500
+      "learning_rate": 3.2601779589402e-05,
+      "loss": 2.9128,
+      "step": 105000
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.258082248501701e-05,
-      "loss": 2.7766,
-      "step": 97600
+      "learning_rate": 3.258520985567762e-05,
+      "loss": 2.906,
+      "step": 105100
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.256297496707133e-05,
-      "loss": 2.7816,
-      "step": 97700
+      "learning_rate": 3.2568640121953244e-05,
+      "loss": 2.9051,
+      "step": 105200
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.254512744912565e-05,
-      "loss": 2.774,
-      "step": 97800
+      "learning_rate": 3.255207038822886e-05,
+      "loss": 2.9118,
+      "step": 105300
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.252727993117997e-05,
-      "loss": 2.7819,
-      "step": 97900
+      "learning_rate": 3.2535500654504487e-05,
+      "loss": 2.92,
+      "step": 105400
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.250943241323429e-05,
-      "loss": 2.7736,
-      "step": 98000
+      "learning_rate": 3.251893092078011e-05,
+      "loss": 2.9047,
+      "step": 105500
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.249158489528861e-05,
-      "loss": 2.7824,
-      "step": 98100
+      "learning_rate": 3.250236118705572e-05,
+      "loss": 2.9151,
+      "step": 105600
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2473737377342934e-05,
-      "loss": 2.7786,
-      "step": 98200
+      "learning_rate": 3.248579145333135e-05,
+      "loss": 2.9157,
+      "step": 105700
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2455889859397255e-05,
-      "loss": 2.7774,
-      "step": 98300
+      "learning_rate": 3.2469221719606966e-05,
+      "loss": 2.9017,
+      "step": 105800
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2438042341451576e-05,
-      "loss": 2.7831,
-      "step": 98400
+      "learning_rate": 3.245265198588259e-05,
+      "loss": 2.9141,
+      "step": 105900
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.24201948235059e-05,
-      "loss": 2.778,
-      "step": 98500
+      "learning_rate": 3.243608225215821e-05,
+      "loss": 2.9044,
+      "step": 106000
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.240234730556022e-05,
-      "loss": 2.7847,
-      "step": 98600
+      "learning_rate": 3.241951251843383e-05,
+      "loss": 2.9065,
+      "step": 106100
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.238449978761453e-05,
-      "loss": 2.7897,
-      "step": 98700
+      "learning_rate": 3.240294278470945e-05,
+      "loss": 2.904,
+      "step": 106200
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2366652269668855e-05,
-      "loss": 2.7819,
-      "step": 98800
+      "learning_rate": 3.238637305098507e-05,
+      "loss": 2.9205,
+      "step": 106300
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.2348804751723176e-05,
-      "loss": 2.777,
-      "step": 98900
+      "learning_rate": 3.2369803317260694e-05,
+      "loss": 2.9195,
+      "step": 106400
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.23309572337775e-05,
-      "loss": 2.7767,
-      "step": 99000
+      "learning_rate": 3.2353233583536316e-05,
+      "loss": 2.9244,
+      "step": 106500
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.231310971583182e-05,
-      "loss": 2.7747,
-      "step": 99100
+      "learning_rate": 3.233666384981193e-05,
+      "loss": 2.9077,
+      "step": 106600
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.229526219788614e-05,
-      "loss": 2.7872,
-      "step": 99200
+      "learning_rate": 3.232009411608756e-05,
+      "loss": 2.9075,
+      "step": 106700
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.227741467994046e-05,
-      "loss": 2.7743,
-      "step": 99300
+      "learning_rate": 3.2303524382363174e-05,
+      "loss": 2.9174,
+      "step": 106800
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.225956716199478e-05,
-      "loss": 2.7711,
-      "step": 99400
+      "learning_rate": 3.2286954648638795e-05,
+      "loss": 2.8982,
+      "step": 106900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.227038491491442e-05,
+      "loss": 2.9094,
+      "step": 107000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.225381518119004e-05,
+      "loss": 2.8942,
+      "step": 107100
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.22417196440491e-05,
-      "loss": 2.7764,
-      "step": 99500
+      "learning_rate": 3.223724544746566e-05,
+      "loss": 2.9079,
+      "step": 107200
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2223872126103424e-05,
-      "loss": 2.7801,
-      "step": 99600
+      "learning_rate": 3.222067571374128e-05,
+      "loss": 2.8946,
+      "step": 107300
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2206024608157745e-05,
-      "loss": 2.7806,
-      "step": 99700
+      "learning_rate": 3.22041059800169e-05,
+      "loss": 2.9088,
+      "step": 107400
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.218817709021207e-05,
-      "loss": 2.7855,
-      "step": 99800
+      "learning_rate": 3.2187536246292524e-05,
+      "loss": 2.9133,
+      "step": 107500
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.217032957226639e-05,
-      "loss": 2.7787,
-      "step": 99900
+      "learning_rate": 3.2170966512568145e-05,
+      "loss": 2.9161,
+      "step": 107600
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.215248205432071e-05,
-      "loss": 2.7816,
-      "step": 100000
+      "learning_rate": 3.215439677884377e-05,
+      "loss": 2.9125,
+      "step": 107700
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.213463453637503e-05,
-      "loss": 2.7644,
-      "step": 100100
+      "learning_rate": 3.213782704511938e-05,
+      "loss": 2.9136,
+      "step": 107800
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.211678701842935e-05,
-      "loss": 2.7853,
-      "step": 100200
+      "learning_rate": 3.212125731139501e-05,
+      "loss": 2.9115,
+      "step": 107900
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.209893950048367e-05,
-      "loss": 2.7787,
-      "step": 100300
+      "learning_rate": 3.210468757767063e-05,
+      "loss": 2.9035,
+      "step": 108000
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2081091982537994e-05,
-      "loss": 2.7706,
-      "step": 100400
+      "learning_rate": 3.2088117843946246e-05,
+      "loss": 2.9133,
+      "step": 108100
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2063244464592315e-05,
-      "loss": 2.7756,
-      "step": 100500
+      "learning_rate": 3.2071548110221874e-05,
+      "loss": 2.9093,
+      "step": 108200
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2045396946646636e-05,
-      "loss": 2.7726,
-      "step": 100600
+      "learning_rate": 3.205497837649749e-05,
+      "loss": 2.9076,
+      "step": 108300
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.202754942870095e-05,
-      "loss": 2.7702,
-      "step": 100700
+      "learning_rate": 3.203840864277311e-05,
+      "loss": 2.9183,
+      "step": 108400
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.200970191075527e-05,
-      "loss": 2.7696,
-      "step": 100800
+      "learning_rate": 3.202183890904873e-05,
+      "loss": 2.9112,
+      "step": 108500
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.199185439280959e-05,
-      "loss": 2.7758,
-      "step": 100900
+      "learning_rate": 3.200526917532435e-05,
+      "loss": 2.9106,
+      "step": 108600
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.1974006874863914e-05,
-      "loss": 2.7863,
-      "step": 101000
+      "learning_rate": 3.1988699441599975e-05,
+      "loss": 2.8978,
+      "step": 108700
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.1956159356918236e-05,
-      "loss": 2.7811,
-      "step": 101100
+      "learning_rate": 3.1972129707875596e-05,
+      "loss": 2.9023,
+      "step": 108800
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.193831183897256e-05,
-      "loss": 2.7844,
-      "step": 101200
+      "learning_rate": 3.195555997415122e-05,
+      "loss": 2.8986,
+      "step": 108900
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.192046432102688e-05,
-      "loss": 2.781,
-      "step": 101300
+      "learning_rate": 3.193899024042684e-05,
+      "loss": 2.8978,
+      "step": 109000
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.19026168030812e-05,
-      "loss": 2.7812,
-      "step": 101400
+      "learning_rate": 3.192242050670246e-05,
+      "loss": 2.9208,
+      "step": 109100
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.188476928513552e-05,
-      "loss": 2.7776,
-      "step": 101500
+      "learning_rate": 3.190585077297808e-05,
+      "loss": 2.9058,
+      "step": 109200
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.186692176718984e-05,
-      "loss": 2.7818,
-      "step": 101600
+      "learning_rate": 3.18892810392537e-05,
+      "loss": 2.914,
+      "step": 109300
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.184907424924416e-05,
-      "loss": 2.7723,
-      "step": 101700
+      "learning_rate": 3.1872711305529325e-05,
+      "loss": 2.9019,
+      "step": 109400
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.183122673129848e-05,
-      "loss": 2.7708,
-      "step": 101800
+      "learning_rate": 3.185614157180494e-05,
+      "loss": 2.9138,
+      "step": 109500
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.18133792133528e-05,
-      "loss": 2.7878,
-      "step": 101900
+      "learning_rate": 3.183957183808056e-05,
+      "loss": 2.9171,
+      "step": 109600
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.179553169540712e-05,
-      "loss": 2.7813,
-      "step": 102000
+      "learning_rate": 3.182300210435619e-05,
+      "loss": 2.8968,
+      "step": 109700
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.177768417746144e-05,
-      "loss": 2.7802,
-      "step": 102100
+      "learning_rate": 3.1806432370631804e-05,
+      "loss": 2.8971,
+      "step": 109800
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.175983665951576e-05,
-      "loss": 2.773,
-      "step": 102200
+      "learning_rate": 3.1789862636907425e-05,
+      "loss": 2.9039,
+      "step": 109900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.177329290318305e-05,
+      "loss": 2.9175,
+      "step": 110000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.175672316945867e-05,
+      "loss": 2.8989,
+      "step": 110100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.174015343573429e-05,
+      "loss": 2.8913,
+      "step": 110200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.172358370200991e-05,
+      "loss": 2.8949,
+      "step": 110300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.170701396828553e-05,
+      "loss": 2.9094,
+      "step": 110400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.169044423456115e-05,
+      "loss": 2.9048,
+      "step": 110500
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1741989141570083e-05,
-      "loss": 2.7742,
-      "step": 102300
+      "learning_rate": 3.1673874500836776e-05,
+      "loss": 2.9012,
+      "step": 110600
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1724141623624405e-05,
-      "loss": 2.7707,
-      "step": 102400
+      "learning_rate": 3.16573047671124e-05,
+      "loss": 2.8962,
+      "step": 110700
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1706294105678726e-05,
-      "loss": 2.783,
-      "step": 102500
+      "learning_rate": 3.164073503338801e-05,
+      "loss": 2.8888,
+      "step": 110800
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.168844658773305e-05,
-      "loss": 2.7739,
-      "step": 102600
+      "learning_rate": 3.162416529966364e-05,
+      "loss": 2.8976,
+      "step": 110900
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.167059906978737e-05,
-      "loss": 2.7649,
-      "step": 102700
+      "learning_rate": 3.1607595565939255e-05,
+      "loss": 2.9046,
+      "step": 111000
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.165275155184169e-05,
-      "loss": 2.7831,
-      "step": 102800
+      "learning_rate": 3.1591025832214876e-05,
+      "loss": 2.9146,
+      "step": 111100
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1634904033896004e-05,
-      "loss": 2.7838,
-      "step": 102900
+      "learning_rate": 3.1574456098490505e-05,
+      "loss": 2.9072,
+      "step": 111200
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1617056515950325e-05,
-      "loss": 2.7795,
-      "step": 103000
+      "learning_rate": 3.155788636476612e-05,
+      "loss": 2.8983,
+      "step": 111300
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1599208998004646e-05,
-      "loss": 2.7791,
-      "step": 103100
+      "learning_rate": 3.154131663104174e-05,
+      "loss": 2.9128,
+      "step": 111400
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.158136148005897e-05,
-      "loss": 2.7725,
-      "step": 103200
+      "learning_rate": 3.152474689731736e-05,
+      "loss": 2.9049,
+      "step": 111500
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.156351396211329e-05,
-      "loss": 2.7783,
-      "step": 103300
+      "learning_rate": 3.1508177163592984e-05,
+      "loss": 2.9088,
+      "step": 111600
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.154566644416761e-05,
-      "loss": 2.7832,
-      "step": 103400
+      "learning_rate": 3.1491607429868605e-05,
+      "loss": 2.9081,
+      "step": 111700
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.152781892622193e-05,
-      "loss": 2.7692,
-      "step": 103500
+      "learning_rate": 3.147503769614422e-05,
+      "loss": 2.9063,
+      "step": 111800
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.150997140827625e-05,
-      "loss": 2.7771,
-      "step": 103600
+      "learning_rate": 3.145846796241985e-05,
+      "loss": 2.9035,
+      "step": 111900
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1492123890330574e-05,
-      "loss": 2.7737,
-      "step": 103700
+      "learning_rate": 3.144189822869546e-05,
+      "loss": 2.8982,
+      "step": 112000
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1474276372384895e-05,
-      "loss": 2.7709,
-      "step": 103800
+      "learning_rate": 3.1425328494971084e-05,
+      "loss": 2.9112,
+      "step": 112100
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.145642885443921e-05,
-      "loss": 2.768,
-      "step": 103900
+      "learning_rate": 3.140875876124671e-05,
+      "loss": 2.9049,
+      "step": 112200
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.143858133649353e-05,
-      "loss": 2.7733,
-      "step": 104000
+      "learning_rate": 3.139218902752233e-05,
+      "loss": 2.8993,
+      "step": 112300
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.142073381854785e-05,
-      "loss": 2.7802,
-      "step": 104100
+      "learning_rate": 3.137561929379795e-05,
+      "loss": 2.9044,
+      "step": 112400
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.140288630060217e-05,
-      "loss": 2.7794,
-      "step": 104200
+      "learning_rate": 3.135904956007357e-05,
+      "loss": 2.9051,
+      "step": 112500
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1385038782656494e-05,
-      "loss": 2.764,
-      "step": 104300
+      "learning_rate": 3.134247982634919e-05,
+      "loss": 2.9084,
+      "step": 112600
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1367191264710815e-05,
-      "loss": 2.7789,
-      "step": 104400
+      "learning_rate": 3.132591009262481e-05,
+      "loss": 2.9031,
+      "step": 112700
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.134934374676514e-05,
-      "loss": 2.7717,
-      "step": 104500
+      "learning_rate": 3.1309340358900434e-05,
+      "loss": 2.884,
+      "step": 112800
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.133149622881946e-05,
-      "loss": 2.7823,
-      "step": 104600
+      "learning_rate": 3.1292770625176056e-05,
+      "loss": 2.8845,
+      "step": 112900
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.1313648710873786e-05,
-      "loss": 2.7802,
-      "step": 104700
+      "learning_rate": 3.127620089145167e-05,
+      "loss": 2.9034,
+      "step": 113000
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.12958011929281e-05,
-      "loss": 2.771,
-      "step": 104800
+      "learning_rate": 3.12596311577273e-05,
+      "loss": 2.9013,
+      "step": 113100
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.127795367498242e-05,
-      "loss": 2.7752,
-      "step": 104900
+      "epoch": 0.38,
+      "learning_rate": 3.124306142400292e-05,
+      "loss": 2.8993,
+      "step": 113200
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.126010615703674e-05,
-      "loss": 2.7721,
-      "step": 105000
+      "epoch": 0.38,
+      "learning_rate": 3.1226491690278535e-05,
+      "loss": 2.9051,
+      "step": 113300
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.1242258639091064e-05,
-      "loss": 2.7679,
-      "step": 105100
+      "learning_rate": 3.120992195655416e-05,
+      "loss": 2.9009,
+      "step": 113400
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.1224411121145385e-05,
-      "loss": 2.7666,
-      "step": 105200
+      "learning_rate": 3.119335222282978e-05,
+      "loss": 2.8966,
+      "step": 113500
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.1206563603199706e-05,
-      "loss": 2.7789,
-      "step": 105300
+      "learning_rate": 3.11767824891054e-05,
+      "loss": 2.9104,
+      "step": 113600
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.118871608525403e-05,
-      "loss": 2.7649,
-      "step": 105400
+      "learning_rate": 3.116021275538102e-05,
+      "loss": 2.9205,
+      "step": 113700
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.117086856730835e-05,
-      "loss": 2.7786,
-      "step": 105500
+      "learning_rate": 3.114364302165664e-05,
+      "loss": 2.9056,
+      "step": 113800
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.115302104936267e-05,
-      "loss": 2.7672,
-      "step": 105600
+      "learning_rate": 3.1127073287932264e-05,
+      "loss": 2.8992,
+      "step": 113900
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.113517353141699e-05,
-      "loss": 2.772,
-      "step": 105700
+      "learning_rate": 3.1110503554207885e-05,
+      "loss": 2.9105,
+      "step": 114000
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.111732601347131e-05,
-      "loss": 2.7773,
-      "step": 105800
+      "learning_rate": 3.109393382048351e-05,
+      "loss": 2.9023,
+      "step": 114100
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.109947849552563e-05,
-      "loss": 2.7653,
-      "step": 105900
+      "learning_rate": 3.107736408675913e-05,
+      "loss": 2.8955,
+      "step": 114200
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.108163097757995e-05,
-      "loss": 2.7805,
-      "step": 106000
+      "learning_rate": 3.106079435303475e-05,
+      "loss": 2.9085,
+      "step": 114300
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.106378345963427e-05,
-      "loss": 2.7822,
-      "step": 106100
+      "learning_rate": 3.104422461931037e-05,
+      "loss": 2.8911,
+      "step": 114400
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.104593594168859e-05,
-      "loss": 2.7739,
-      "step": 106200
+      "learning_rate": 3.1027654885585986e-05,
+      "loss": 2.9028,
+      "step": 114500
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.102808842374291e-05,
-      "loss": 2.7786,
-      "step": 106300
+      "learning_rate": 3.1011085151861614e-05,
+      "loss": 2.8944,
+      "step": 114600
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.101024090579723e-05,
-      "loss": 2.7724,
-      "step": 106400
+      "learning_rate": 3.099451541813723e-05,
+      "loss": 2.9191,
+      "step": 114700
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.0992393387851554e-05,
-      "loss": 2.7703,
-      "step": 106500
+      "learning_rate": 3.097794568441285e-05,
+      "loss": 2.8898,
+      "step": 114800
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.0974545869905875e-05,
-      "loss": 2.7688,
-      "step": 106600
+      "learning_rate": 3.096137595068848e-05,
+      "loss": 2.9047,
+      "step": 114900
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.0956698351960197e-05,
-      "loss": 2.7611,
-      "step": 106700
+      "learning_rate": 3.094480621696409e-05,
+      "loss": 2.9047,
+      "step": 115000
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.093885083401452e-05,
-      "loss": 2.7784,
-      "step": 106800
+      "learning_rate": 3.0928236483239715e-05,
+      "loss": 2.9011,
+      "step": 115100
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.092100331606884e-05,
-      "loss": 2.7738,
-      "step": 106900
+      "learning_rate": 3.0911666749515336e-05,
+      "loss": 2.8916,
+      "step": 115200
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.0903155798123153e-05,
-      "loss": 2.7673,
-      "step": 107000
+      "learning_rate": 3.089509701579096e-05,
+      "loss": 2.8964,
+      "step": 115300
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.0885308280177475e-05,
-      "loss": 2.7668,
-      "step": 107100
+      "learning_rate": 3.087852728206658e-05,
+      "loss": 2.8854,
+      "step": 115400
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.0867460762231796e-05,
-      "loss": 2.7788,
-      "step": 107200
+      "learning_rate": 3.08619575483422e-05,
+      "loss": 2.8961,
+      "step": 115500
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.084961324428612e-05,
-      "loss": 2.767,
-      "step": 107300
+      "learning_rate": 3.084538781461782e-05,
+      "loss": 2.9032,
+      "step": 115600
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.083176572634044e-05,
-      "loss": 2.7629,
-      "step": 107400
+      "learning_rate": 3.0828818080893443e-05,
+      "loss": 2.9132,
+      "step": 115700
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.081391820839476e-05,
-      "loss": 2.7755,
-      "step": 107500
+      "learning_rate": 3.0812248347169065e-05,
+      "loss": 2.9255,
+      "step": 115800
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.079607069044908e-05,
-      "loss": 2.776,
-      "step": 107600
+      "learning_rate": 3.0795678613444686e-05,
+      "loss": 2.9171,
+      "step": 115900
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.07782231725034e-05,
-      "loss": 2.7787,
-      "step": 107700
+      "learning_rate": 3.07791088797203e-05,
+      "loss": 2.9023,
+      "step": 116000
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.076037565455772e-05,
-      "loss": 2.772,
-      "step": 107800
+      "learning_rate": 3.076253914599593e-05,
+      "loss": 2.9057,
+      "step": 116100
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0742528136612044e-05,
-      "loss": 2.7732,
-      "step": 107900
+      "learning_rate": 3.0745969412271544e-05,
+      "loss": 2.9089,
+      "step": 116200
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0724680618666366e-05,
-      "loss": 2.7695,
-      "step": 108000
+      "learning_rate": 3.0729399678547165e-05,
+      "loss": 2.9149,
+      "step": 116300
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.070683310072068e-05,
-      "loss": 2.7793,
-      "step": 108100
+      "learning_rate": 3.0712829944822794e-05,
+      "loss": 2.8992,
+      "step": 116400
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0688985582775e-05,
-      "loss": 2.7695,
-      "step": 108200
+      "learning_rate": 3.069626021109841e-05,
+      "loss": 2.9032,
+      "step": 116500
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.067113806482932e-05,
-      "loss": 2.7743,
-      "step": 108300
+      "learning_rate": 3.067969047737403e-05,
+      "loss": 2.8852,
+      "step": 116600
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0653290546883644e-05,
-      "loss": 2.7815,
-      "step": 108400
+      "learning_rate": 3.066312074364965e-05,
+      "loss": 2.913,
+      "step": 116700
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0635443028937965e-05,
-      "loss": 2.7798,
-      "step": 108500
+      "learning_rate": 3.064655100992527e-05,
+      "loss": 2.9085,
+      "step": 116800
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0617595510992286e-05,
-      "loss": 2.7774,
-      "step": 108600
+      "learning_rate": 3.0629981276200894e-05,
+      "loss": 2.8964,
+      "step": 116900
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.059974799304661e-05,
-      "loss": 2.7779,
-      "step": 108700
+      "learning_rate": 3.061341154247651e-05,
+      "loss": 2.8906,
+      "step": 117000
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.058190047510093e-05,
-      "loss": 2.7747,
-      "step": 108800
+      "learning_rate": 3.059684180875214e-05,
+      "loss": 2.9057,
+      "step": 117100
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.056405295715525e-05,
-      "loss": 2.7721,
-      "step": 108900
+      "learning_rate": 3.058027207502775e-05,
+      "loss": 2.8965,
+      "step": 117200
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.054620543920957e-05,
-      "loss": 2.7749,
-      "step": 109000
+      "learning_rate": 3.056370234130337e-05,
+      "loss": 2.9067,
+      "step": 117300
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.052835792126389e-05,
-      "loss": 2.767,
-      "step": 109100
+      "learning_rate": 3.0547132607579e-05,
+      "loss": 2.895,
+      "step": 117400
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.051051040331821e-05,
-      "loss": 2.7563,
-      "step": 109200
+      "learning_rate": 3.0530562873854616e-05,
+      "loss": 2.9112,
+      "step": 117500
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.049266288537253e-05,
-      "loss": 2.7759,
-      "step": 109300
+      "learning_rate": 3.0513993140130238e-05,
+      "loss": 2.899,
+      "step": 117600
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0474815367426852e-05,
-      "loss": 2.766,
-      "step": 109400
+      "learning_rate": 3.0497423406405863e-05,
+      "loss": 2.912,
+      "step": 117700
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.045696784948117e-05,
-      "loss": 2.7694,
-      "step": 109500
+      "learning_rate": 3.048085367268148e-05,
+      "loss": 2.902,
+      "step": 117800
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.043912033153549e-05,
-      "loss": 2.7632,
-      "step": 109600
+      "learning_rate": 3.04642839389571e-05,
+      "loss": 2.9101,
+      "step": 117900
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0421272813589813e-05,
-      "loss": 2.7565,
-      "step": 109700
+      "learning_rate": 3.0447714205232724e-05,
+      "loss": 2.8883,
+      "step": 118000
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0403425295644137e-05,
-      "loss": 2.7679,
-      "step": 109800
+      "learning_rate": 3.0431144471508345e-05,
+      "loss": 2.8977,
+      "step": 118100
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.038557777769846e-05,
-      "loss": 2.7621,
-      "step": 109900
+      "learning_rate": 3.0414574737783963e-05,
+      "loss": 2.8954,
+      "step": 118200
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.036773025975278e-05,
-      "loss": 2.7688,
-      "step": 110000
+      "learning_rate": 3.0398005004059588e-05,
+      "loss": 2.8856,
+      "step": 118300
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.03498827418071e-05,
-      "loss": 2.7615,
-      "step": 110100
+      "learning_rate": 3.0381435270335206e-05,
+      "loss": 2.9129,
+      "step": 118400
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0332035223861422e-05,
-      "loss": 2.7697,
-      "step": 110200
+      "learning_rate": 3.0364865536610828e-05,
+      "loss": 2.9026,
+      "step": 118500
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0314187705915743e-05,
-      "loss": 2.7759,
-      "step": 110300
+      "learning_rate": 3.034829580288645e-05,
+      "loss": 2.8977,
+      "step": 118600
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.029634018797006e-05,
-      "loss": 2.7709,
-      "step": 110400
+      "learning_rate": 3.033172606916207e-05,
+      "loss": 2.906,
+      "step": 118700
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0278492670024382e-05,
-      "loss": 2.7642,
-      "step": 110500
+      "learning_rate": 3.031515633543769e-05,
+      "loss": 2.9019,
+      "step": 118800
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.0260645152078704e-05,
-      "loss": 2.764,
-      "step": 110600
+      "learning_rate": 3.0298586601713313e-05,
+      "loss": 2.9118,
+      "step": 118900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.028201686798893e-05,
+      "loss": 2.9018,
+      "step": 119000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0265447134264553e-05,
+      "loss": 2.8995,
+      "step": 119100
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0242797634133025e-05,
-      "loss": 2.7764,
-      "step": 110700
+      "learning_rate": 3.0248877400540178e-05,
+      "loss": 2.8886,
+      "step": 119200
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0224950116187346e-05,
-      "loss": 2.7683,
-      "step": 110800
+      "learning_rate": 3.0232307666815796e-05,
+      "loss": 2.9048,
+      "step": 119300
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0207102598241664e-05,
-      "loss": 2.7657,
-      "step": 110900
+      "learning_rate": 3.0215737933091414e-05,
+      "loss": 2.9034,
+      "step": 119400
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0189255080295985e-05,
-      "loss": 2.7731,
-      "step": 111000
+      "learning_rate": 3.019916819936704e-05,
+      "loss": 2.903,
+      "step": 119500
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0171407562350306e-05,
-      "loss": 2.7703,
-      "step": 111100
+      "learning_rate": 3.0182598465642657e-05,
+      "loss": 2.9042,
+      "step": 119600
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0153560044404628e-05,
-      "loss": 2.7718,
-      "step": 111200
+      "learning_rate": 3.016602873191828e-05,
+      "loss": 2.8954,
+      "step": 119700
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.013571252645895e-05,
-      "loss": 2.7739,
-      "step": 111300
+      "learning_rate": 3.0149458998193903e-05,
+      "loss": 2.9163,
+      "step": 119800
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.011786500851327e-05,
-      "loss": 2.7715,
-      "step": 111400
+      "learning_rate": 3.013288926446952e-05,
+      "loss": 2.904,
+      "step": 119900
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0100017490567588e-05,
-      "loss": 2.7594,
-      "step": 111500
+      "learning_rate": 3.011631953074514e-05,
+      "loss": 2.9007,
+      "step": 120000
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.008216997262191e-05,
-      "loss": 2.7705,
-      "step": 111600
+      "learning_rate": 3.0099749797020764e-05,
+      "loss": 2.9006,
+      "step": 120100
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.006432245467623e-05,
-      "loss": 2.7714,
-      "step": 111700
+      "learning_rate": 3.0083180063296386e-05,
+      "loss": 2.8923,
+      "step": 120200
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.004647493673055e-05,
-      "loss": 2.7601,
-      "step": 111800
+      "learning_rate": 3.0066610329572004e-05,
+      "loss": 2.8989,
+      "step": 120300
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.0028627418784873e-05,
-      "loss": 2.7659,
-      "step": 111900
+      "learning_rate": 3.005004059584763e-05,
+      "loss": 2.9075,
+      "step": 120400
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.001077990083919e-05,
-      "loss": 2.7733,
-      "step": 112000
+      "learning_rate": 3.0033470862123247e-05,
+      "loss": 2.8959,
+      "step": 120500
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9992932382893512e-05,
-      "loss": 2.7685,
-      "step": 112100
+      "learning_rate": 3.0016901128398868e-05,
+      "loss": 2.9021,
+      "step": 120600
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9975084864947833e-05,
-      "loss": 2.7684,
-      "step": 112200
+      "learning_rate": 3.000033139467449e-05,
+      "loss": 2.904,
+      "step": 120700
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9957237347002154e-05,
-      "loss": 2.7722,
-      "step": 112300
+      "learning_rate": 2.998376166095011e-05,
+      "loss": 2.8865,
+      "step": 120800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.996719192722573e-05,
+      "loss": 2.8898,
+      "step": 120900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9950622193501354e-05,
+      "loss": 2.8942,
+      "step": 121000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9934052459776972e-05,
+      "loss": 2.9035,
+      "step": 121100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9917482726052594e-05,
+      "loss": 2.9038,
+      "step": 121200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.990091299232822e-05,
+      "loss": 2.8879,
+      "step": 121300
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9939389829056475e-05,
-      "loss": 2.7639,
-      "step": 112400
+      "learning_rate": 2.9884343258603837e-05,
+      "loss": 2.8921,
+      "step": 121400
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9921542311110797e-05,
-      "loss": 2.7693,
-      "step": 112500
+      "learning_rate": 2.9867773524879455e-05,
+      "loss": 2.8896,
+      "step": 121500
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9903694793165114e-05,
-      "loss": 2.759,
-      "step": 112600
+      "learning_rate": 2.985120379115508e-05,
+      "loss": 2.8892,
+      "step": 121600
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9885847275219436e-05,
-      "loss": 2.7726,
-      "step": 112700
+      "learning_rate": 2.98346340574307e-05,
+      "loss": 2.8968,
+      "step": 121700
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9867999757273757e-05,
-      "loss": 2.7667,
-      "step": 112800
+      "learning_rate": 2.981806432370632e-05,
+      "loss": 2.8922,
+      "step": 121800
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9850152239328078e-05,
-      "loss": 2.7621,
-      "step": 112900
+      "learning_rate": 2.9801494589981944e-05,
+      "loss": 2.8962,
+      "step": 121900
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.98323047213824e-05,
-      "loss": 2.772,
-      "step": 113000
+      "learning_rate": 2.9784924856257562e-05,
+      "loss": 2.9117,
+      "step": 122000
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9814457203436717e-05,
-      "loss": 2.7647,
-      "step": 113100
+      "learning_rate": 2.976835512253318e-05,
+      "loss": 2.8963,
+      "step": 122100
     },
     {
       "epoch": 0.4,
-      "learning_rate": 2.9796609685491038e-05,
-      "loss": 2.7604,
-      "step": 113200
+      "learning_rate": 2.97517853888088e-05,
+      "loss": 2.8966,
+      "step": 122200
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.977876216754536e-05,
-      "loss": 2.7798,
-      "step": 113300
+      "epoch": 0.41,
+      "learning_rate": 2.9735215655084426e-05,
+      "loss": 2.9019,
+      "step": 122300
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 2.976091464959968e-05,
-      "loss": 2.7606,
-      "step": 113400
+      "epoch": 0.41,
+      "learning_rate": 2.9718645921360044e-05,
+      "loss": 2.9001,
+      "step": 122400
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9743067131654002e-05,
-      "loss": 2.7628,
-      "step": 113500
+      "learning_rate": 2.9702076187635663e-05,
+      "loss": 2.8922,
+      "step": 122500
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9725219613708323e-05,
-      "loss": 2.7717,
-      "step": 113600
+      "learning_rate": 2.9685506453911287e-05,
+      "loss": 2.9061,
+      "step": 122600
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.970737209576264e-05,
-      "loss": 2.7719,
-      "step": 113700
+      "learning_rate": 2.966893672018691e-05,
+      "loss": 2.8953,
+      "step": 122700
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9689524577816962e-05,
-      "loss": 2.7632,
-      "step": 113800
+      "learning_rate": 2.9652366986462527e-05,
+      "loss": 2.9,
+      "step": 122800
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9671677059871283e-05,
-      "loss": 2.7752,
-      "step": 113900
+      "learning_rate": 2.9635797252738152e-05,
+      "loss": 2.8986,
+      "step": 122900
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9653829541925605e-05,
-      "loss": 2.7617,
-      "step": 114000
+      "learning_rate": 2.961922751901377e-05,
+      "loss": 2.8879,
+      "step": 123000
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9635982023979926e-05,
-      "loss": 2.7669,
-      "step": 114100
+      "learning_rate": 2.9602657785289388e-05,
+      "loss": 2.9089,
+      "step": 123100
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9618134506034244e-05,
-      "loss": 2.7616,
-      "step": 114200
+      "learning_rate": 2.9586088051565013e-05,
+      "loss": 2.8993,
+      "step": 123200
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9600286988088565e-05,
-      "loss": 2.7716,
-      "step": 114300
+      "learning_rate": 2.9569518317840634e-05,
+      "loss": 2.8943,
+      "step": 123300
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9582439470142886e-05,
-      "loss": 2.7509,
-      "step": 114400
+      "learning_rate": 2.9552948584116252e-05,
+      "loss": 2.9056,
+      "step": 123400
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9564591952197207e-05,
-      "loss": 2.7704,
-      "step": 114500
+      "learning_rate": 2.9536378850391877e-05,
+      "loss": 2.9072,
+      "step": 123500
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.954674443425153e-05,
-      "loss": 2.7653,
-      "step": 114600
+      "learning_rate": 2.9519809116667495e-05,
+      "loss": 2.8973,
+      "step": 123600
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.952889691630585e-05,
-      "loss": 2.7519,
-      "step": 114700
+      "learning_rate": 2.9503239382943117e-05,
+      "loss": 2.8908,
+      "step": 123700
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9511049398360168e-05,
-      "loss": 2.7644,
-      "step": 114800
+      "learning_rate": 2.948666964921874e-05,
+      "loss": 2.8968,
+      "step": 123800
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9493201880414496e-05,
-      "loss": 2.7755,
-      "step": 114900
+      "learning_rate": 2.947009991549436e-05,
+      "loss": 2.8994,
+      "step": 123900
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9475354362468817e-05,
-      "loss": 2.7644,
-      "step": 115000
+      "learning_rate": 2.9453530181769978e-05,
+      "loss": 2.9112,
+      "step": 124000
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9457506844523135e-05,
-      "loss": 2.7642,
-      "step": 115100
+      "learning_rate": 2.9436960448045603e-05,
+      "loss": 2.9016,
+      "step": 124100
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9439659326577456e-05,
-      "loss": 2.7685,
-      "step": 115200
+      "learning_rate": 2.942039071432122e-05,
+      "loss": 2.9056,
+      "step": 124200
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9421811808631777e-05,
-      "loss": 2.7611,
-      "step": 115300
+      "learning_rate": 2.9403820980596842e-05,
+      "loss": 2.8962,
+      "step": 124300
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9403964290686098e-05,
-      "loss": 2.7727,
-      "step": 115400
+      "learning_rate": 2.9387251246872467e-05,
+      "loss": 2.8942,
+      "step": 124400
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.938611677274042e-05,
-      "loss": 2.7658,
-      "step": 115500
+      "learning_rate": 2.9370681513148085e-05,
+      "loss": 2.8965,
+      "step": 124500
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.936826925479474e-05,
-      "loss": 2.7597,
-      "step": 115600
+      "learning_rate": 2.9354111779423703e-05,
+      "loss": 2.8933,
+      "step": 124600
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.935042173684906e-05,
-      "loss": 2.761,
-      "step": 115700
+      "learning_rate": 2.9337542045699328e-05,
+      "loss": 2.8922,
+      "step": 124700
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.933257421890338e-05,
-      "loss": 2.7679,
-      "step": 115800
+      "learning_rate": 2.932097231197495e-05,
+      "loss": 2.8975,
+      "step": 124800
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.93147267009577e-05,
-      "loss": 2.7595,
-      "step": 115900
+      "learning_rate": 2.9304402578250568e-05,
+      "loss": 2.8984,
+      "step": 124900
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9296879183012022e-05,
-      "loss": 2.7625,
-      "step": 116000
+      "learning_rate": 2.9287832844526192e-05,
+      "loss": 2.9095,
+      "step": 125000
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.9279031665066343e-05,
-      "loss": 2.7687,
-      "step": 116100
+      "learning_rate": 2.927126311080181e-05,
+      "loss": 2.8802,
+      "step": 125100
     },
     {
       "epoch": 0.41,
-      "learning_rate": 2.926118414712066e-05,
-      "loss": 2.7706,
-      "step": 116200
+      "learning_rate": 2.925469337707743e-05,
+      "loss": 2.9026,
+      "step": 125200
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9243336629174982e-05,
-      "loss": 2.7599,
-      "step": 116300
+      "learning_rate": 2.9238123643353053e-05,
+      "loss": 2.895,
+      "step": 125300
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9225489111229304e-05,
-      "loss": 2.7678,
-      "step": 116400
+      "learning_rate": 2.9221553909628675e-05,
+      "loss": 2.8923,
+      "step": 125400
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9207641593283625e-05,
-      "loss": 2.7669,
-      "step": 116500
+      "learning_rate": 2.9204984175904293e-05,
+      "loss": 2.8907,
+      "step": 125500
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9189794075337946e-05,
-      "loss": 2.7563,
-      "step": 116600
+      "learning_rate": 2.9188414442179918e-05,
+      "loss": 2.8988,
+      "step": 125600
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9171946557392267e-05,
-      "loss": 2.7615,
-      "step": 116700
+      "learning_rate": 2.9171844708455536e-05,
+      "loss": 2.8979,
+      "step": 125700
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9154099039446585e-05,
-      "loss": 2.7517,
-      "step": 116800
+      "learning_rate": 2.9155274974731157e-05,
+      "loss": 2.9,
+      "step": 125800
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9136251521500906e-05,
-      "loss": 2.7608,
-      "step": 116900
+      "learning_rate": 2.9138705241006782e-05,
+      "loss": 2.9035,
+      "step": 125900
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9118404003555228e-05,
-      "loss": 2.769,
-      "step": 117000
+      "learning_rate": 2.91221355072824e-05,
+      "loss": 2.9008,
+      "step": 126000
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.910055648560955e-05,
-      "loss": 2.7616,
-      "step": 117100
+      "learning_rate": 2.910556577355802e-05,
+      "loss": 2.9049,
+      "step": 126100
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.908270896766387e-05,
-      "loss": 2.7562,
-      "step": 117200
+      "learning_rate": 2.9088996039833643e-05,
+      "loss": 2.8749,
+      "step": 126200
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.9064861449718188e-05,
-      "loss": 2.7626,
-      "step": 117300
+      "learning_rate": 2.907242630610926e-05,
+      "loss": 2.9003,
+      "step": 126300
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.904701393177251e-05,
-      "loss": 2.7589,
-      "step": 117400
+      "learning_rate": 2.9055856572384883e-05,
+      "loss": 2.903,
+      "step": 126400
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.902916641382683e-05,
-      "loss": 2.7657,
-      "step": 117500
+      "learning_rate": 2.9039286838660508e-05,
+      "loss": 2.9018,
+      "step": 126500
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.901131889588115e-05,
-      "loss": 2.7595,
-      "step": 117600
+      "learning_rate": 2.9022717104936126e-05,
+      "loss": 2.9062,
+      "step": 126600
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8993471377935473e-05,
-      "loss": 2.7686,
-      "step": 117700
+      "learning_rate": 2.9006147371211744e-05,
+      "loss": 2.8852,
+      "step": 126700
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8975623859989794e-05,
-      "loss": 2.7624,
-      "step": 117800
+      "learning_rate": 2.898957763748737e-05,
+      "loss": 2.8955,
+      "step": 126800
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.895777634204411e-05,
-      "loss": 2.7675,
-      "step": 117900
+      "learning_rate": 2.897300790376299e-05,
+      "loss": 2.8958,
+      "step": 126900
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8939928824098433e-05,
-      "loss": 2.759,
-      "step": 118000
+      "learning_rate": 2.8956438170038608e-05,
+      "loss": 2.9063,
+      "step": 127000
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8922081306152754e-05,
-      "loss": 2.7601,
-      "step": 118100
+      "learning_rate": 2.8939868436314233e-05,
+      "loss": 2.9065,
+      "step": 127100
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8904233788207075e-05,
-      "loss": 2.7665,
-      "step": 118200
+      "learning_rate": 2.892329870258985e-05,
+      "loss": 2.8945,
+      "step": 127200
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8886386270261397e-05,
-      "loss": 2.7626,
-      "step": 118300
+      "learning_rate": 2.890672896886547e-05,
+      "loss": 2.8994,
+      "step": 127300
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8868538752315714e-05,
-      "loss": 2.7513,
-      "step": 118400
+      "learning_rate": 2.889015923514109e-05,
+      "loss": 2.888,
+      "step": 127400
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8850691234370036e-05,
-      "loss": 2.7645,
-      "step": 118500
+      "learning_rate": 2.8873589501416716e-05,
+      "loss": 2.8907,
+      "step": 127500
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8832843716424357e-05,
-      "loss": 2.7585,
-      "step": 118600
+      "learning_rate": 2.8857019767692334e-05,
+      "loss": 2.8881,
+      "step": 127600
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8814996198478678e-05,
-      "loss": 2.7556,
-      "step": 118700
+      "learning_rate": 2.8840450033967952e-05,
+      "loss": 2.8798,
+      "step": 127700
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8797148680533e-05,
-      "loss": 2.7694,
-      "step": 118800
+      "learning_rate": 2.8823880300243577e-05,
+      "loss": 2.8915,
+      "step": 127800
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.877930116258732e-05,
-      "loss": 2.7676,
-      "step": 118900
+      "learning_rate": 2.8807310566519198e-05,
+      "loss": 2.9,
+      "step": 127900
     },
     {
       "epoch": 0.42,
-      "learning_rate": 2.8761453644641638e-05,
-      "loss": 2.7649,
-      "step": 119000
+      "learning_rate": 2.8790740832794816e-05,
+      "loss": 2.9083,
+      "step": 128000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.877417109907044e-05,
+      "loss": 2.8807,
+      "step": 128100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.875760136534606e-05,
+      "loss": 2.8994,
+      "step": 128200
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.874360612669596e-05,
-      "loss": 2.7552,
-      "step": 119100
+      "learning_rate": 2.8741031631621677e-05,
+      "loss": 2.895,
+      "step": 128300
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.872575860875028e-05,
-      "loss": 2.7612,
-      "step": 119200
+      "learning_rate": 2.8724461897897302e-05,
+      "loss": 2.8952,
+      "step": 128400
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8707911090804602e-05,
-      "loss": 2.7602,
-      "step": 119300
+      "learning_rate": 2.8707892164172923e-05,
+      "loss": 2.8965,
+      "step": 128500
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8690063572858923e-05,
-      "loss": 2.7551,
-      "step": 119400
+      "learning_rate": 2.869132243044854e-05,
+      "loss": 2.9009,
+      "step": 128600
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.867221605491324e-05,
-      "loss": 2.7676,
-      "step": 119500
+      "learning_rate": 2.8674752696724166e-05,
+      "loss": 2.896,
+      "step": 128700
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8654368536967562e-05,
-      "loss": 2.7624,
-      "step": 119600
+      "learning_rate": 2.8658182962999784e-05,
+      "loss": 2.9016,
+      "step": 128800
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8636521019021883e-05,
-      "loss": 2.7723,
-      "step": 119700
+      "learning_rate": 2.8641613229275406e-05,
+      "loss": 2.8877,
+      "step": 128900
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8618673501076205e-05,
-      "loss": 2.754,
-      "step": 119800
+      "learning_rate": 2.862504349555103e-05,
+      "loss": 2.8922,
+      "step": 129000
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8600825983130526e-05,
-      "loss": 2.763,
-      "step": 119900
+      "learning_rate": 2.860847376182665e-05,
+      "loss": 2.9068,
+      "step": 129100
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8582978465184847e-05,
-      "loss": 2.7562,
-      "step": 120000
+      "learning_rate": 2.8591904028102267e-05,
+      "loss": 2.8866,
+      "step": 129200
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.856513094723917e-05,
-      "loss": 2.7504,
-      "step": 120100
+      "learning_rate": 2.8575334294377892e-05,
+      "loss": 2.8893,
+      "step": 129300
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8547283429293493e-05,
-      "loss": 2.7601,
-      "step": 120200
+      "learning_rate": 2.855876456065351e-05,
+      "loss": 2.8973,
+      "step": 129400
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8529435911347814e-05,
-      "loss": 2.7528,
-      "step": 120300
+      "learning_rate": 2.854219482692913e-05,
+      "loss": 2.8843,
+      "step": 129500
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8511588393402132e-05,
-      "loss": 2.7636,
-      "step": 120400
+      "learning_rate": 2.8525625093204756e-05,
+      "loss": 2.8933,
+      "step": 129600
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8493740875456453e-05,
-      "loss": 2.7589,
-      "step": 120500
+      "learning_rate": 2.8509055359480374e-05,
+      "loss": 2.9022,
+      "step": 129700
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8475893357510774e-05,
-      "loss": 2.7598,
-      "step": 120600
+      "learning_rate": 2.8492485625755992e-05,
+      "loss": 2.9046,
+      "step": 129800
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8458045839565096e-05,
-      "loss": 2.7556,
-      "step": 120700
+      "learning_rate": 2.8475915892031617e-05,
+      "loss": 2.885,
+      "step": 129900
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8440198321619417e-05,
-      "loss": 2.757,
-      "step": 120800
+      "learning_rate": 2.845934615830724e-05,
+      "loss": 2.8884,
+      "step": 130000
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8422350803673735e-05,
-      "loss": 2.7574,
-      "step": 120900
+      "learning_rate": 2.8442776424582857e-05,
+      "loss": 2.9042,
+      "step": 130100
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8404503285728056e-05,
-      "loss": 2.7673,
-      "step": 121000
+      "learning_rate": 2.842620669085848e-05,
+      "loss": 2.9146,
+      "step": 130200
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8386655767782377e-05,
-      "loss": 2.7559,
-      "step": 121100
+      "learning_rate": 2.84096369571341e-05,
+      "loss": 2.8692,
+      "step": 130300
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8368808249836698e-05,
-      "loss": 2.7568,
-      "step": 121200
+      "learning_rate": 2.839306722340972e-05,
+      "loss": 2.883,
+      "step": 130400
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.835096073189102e-05,
-      "loss": 2.7604,
-      "step": 121300
+      "learning_rate": 2.8376497489685343e-05,
+      "loss": 2.8774,
+      "step": 130500
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.833311321394534e-05,
-      "loss": 2.7564,
-      "step": 121400
+      "learning_rate": 2.8359927755960964e-05,
+      "loss": 2.8901,
+      "step": 130600
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.831526569599966e-05,
-      "loss": 2.7626,
-      "step": 121500
+      "learning_rate": 2.8343358022236582e-05,
+      "loss": 2.876,
+      "step": 130700
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.829741817805398e-05,
-      "loss": 2.7636,
-      "step": 121600
+      "learning_rate": 2.8326788288512207e-05,
+      "loss": 2.9048,
+      "step": 130800
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.82795706601083e-05,
-      "loss": 2.7605,
-      "step": 121700
+      "learning_rate": 2.8310218554787825e-05,
+      "loss": 2.8872,
+      "step": 130900
     },
     {
       "epoch": 0.43,
-      "learning_rate": 2.8261723142162622e-05,
-      "loss": 2.7583,
-      "step": 121800
+      "learning_rate": 2.8293648821063447e-05,
+      "loss": 2.8883,
+      "step": 131000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.827707908733907e-05,
+      "loss": 2.8979,
+      "step": 131100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.826050935361469e-05,
+      "loss": 2.8869,
+      "step": 131200
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8243875624216943e-05,
-      "loss": 2.7727,
-      "step": 121900
+      "learning_rate": 2.8243939619890308e-05,
+      "loss": 2.891,
+      "step": 131300
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.822602810627126e-05,
-      "loss": 2.7519,
-      "step": 122000
+      "learning_rate": 2.8227369886165932e-05,
+      "loss": 2.874,
+      "step": 131400
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8208180588325582e-05,
-      "loss": 2.7585,
-      "step": 122100
+      "learning_rate": 2.821080015244155e-05,
+      "loss": 2.8911,
+      "step": 131500
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8190333070379904e-05,
-      "loss": 2.7645,
-      "step": 122200
+      "learning_rate": 2.8194230418717172e-05,
+      "loss": 2.8811,
+      "step": 131600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8177660684992797e-05,
+      "loss": 2.8937,
+      "step": 131700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8161090951268415e-05,
+      "loss": 2.8806,
+      "step": 131800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8144521217544033e-05,
+      "loss": 2.9024,
+      "step": 131900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8127951483819658e-05,
+      "loss": 2.8879,
+      "step": 132000
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8172485552434225e-05,
-      "loss": 2.7689,
-      "step": 122300
+      "learning_rate": 2.811138175009528e-05,
+      "loss": 2.8792,
+      "step": 132100
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8154638034488546e-05,
-      "loss": 2.7523,
-      "step": 122400
+      "learning_rate": 2.8094812016370897e-05,
+      "loss": 2.9023,
+      "step": 132200
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8136790516542867e-05,
-      "loss": 2.7602,
-      "step": 122500
+      "learning_rate": 2.8078242282646522e-05,
+      "loss": 2.8924,
+      "step": 132300
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8118942998597185e-05,
-      "loss": 2.756,
-      "step": 122600
+      "learning_rate": 2.806167254892214e-05,
+      "loss": 2.9017,
+      "step": 132400
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8101095480651506e-05,
-      "loss": 2.7626,
-      "step": 122700
+      "learning_rate": 2.8045102815197762e-05,
+      "loss": 2.892,
+      "step": 132500
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8083247962705828e-05,
-      "loss": 2.7616,
-      "step": 122800
+      "learning_rate": 2.802853308147338e-05,
+      "loss": 2.8887,
+      "step": 132600
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.806540044476015e-05,
-      "loss": 2.7545,
-      "step": 122900
+      "learning_rate": 2.8011963347749005e-05,
+      "loss": 2.8902,
+      "step": 132700
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.804755292681447e-05,
-      "loss": 2.7735,
-      "step": 123000
+      "learning_rate": 2.7995393614024623e-05,
+      "loss": 2.8983,
+      "step": 132800
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.8029705408868788e-05,
-      "loss": 2.761,
-      "step": 123100
+      "learning_rate": 2.797882388030024e-05,
+      "loss": 2.8911,
+      "step": 132900
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.801185789092311e-05,
-      "loss": 2.7621,
-      "step": 123200
+      "learning_rate": 2.7962254146575866e-05,
+      "loss": 2.8953,
+      "step": 133000
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.799401037297743e-05,
-      "loss": 2.7577,
-      "step": 123300
+      "learning_rate": 2.7945684412851487e-05,
+      "loss": 2.8784,
+      "step": 133100
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.797616285503175e-05,
-      "loss": 2.7498,
-      "step": 123400
+      "learning_rate": 2.7929114679127105e-05,
+      "loss": 2.9099,
+      "step": 133200
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7958315337086073e-05,
-      "loss": 2.7634,
-      "step": 123500
+      "learning_rate": 2.791254494540273e-05,
+      "loss": 2.903,
+      "step": 133300
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7940467819140394e-05,
-      "loss": 2.764,
-      "step": 123600
+      "learning_rate": 2.7895975211678348e-05,
+      "loss": 2.8903,
+      "step": 133400
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.792262030119471e-05,
-      "loss": 2.7741,
-      "step": 123700
+      "learning_rate": 2.787940547795397e-05,
+      "loss": 2.8966,
+      "step": 133500
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7904772783249033e-05,
-      "loss": 2.7674,
-      "step": 123800
+      "learning_rate": 2.786283574422959e-05,
+      "loss": 2.8818,
+      "step": 133600
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7886925265303354e-05,
-      "loss": 2.748,
-      "step": 123900
+      "learning_rate": 2.7846266010505213e-05,
+      "loss": 2.8896,
+      "step": 133700
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7869077747357675e-05,
-      "loss": 2.7595,
-      "step": 124000
+      "learning_rate": 2.782969627678083e-05,
+      "loss": 2.8928,
+      "step": 133800
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7851230229411997e-05,
-      "loss": 2.7542,
-      "step": 124100
+      "learning_rate": 2.7813126543056456e-05,
+      "loss": 2.8852,
+      "step": 133900
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7833382711466314e-05,
-      "loss": 2.7677,
-      "step": 124200
+      "learning_rate": 2.7796556809332074e-05,
+      "loss": 2.8965,
+      "step": 134000
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7815535193520636e-05,
-      "loss": 2.7678,
-      "step": 124300
+      "learning_rate": 2.7779987075607695e-05,
+      "loss": 2.8829,
+      "step": 134100
     },
     {
       "epoch": 0.44,
-      "learning_rate": 2.7797687675574957e-05,
-      "loss": 2.7632,
-      "step": 124400
+      "learning_rate": 2.776341734188332e-05,
+      "loss": 2.8979,
+      "step": 134200
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.7779840157629278e-05,
-      "loss": 2.7536,
-      "step": 124500
+      "epoch": 0.45,
+      "learning_rate": 2.7746847608158938e-05,
+      "loss": 2.9022,
+      "step": 134300
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 2.77619926396836e-05,
-      "loss": 2.7657,
-      "step": 124600
+      "epoch": 0.45,
+      "learning_rate": 2.7730277874434556e-05,
+      "loss": 2.9065,
+      "step": 134400
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.774414512173792e-05,
-      "loss": 2.7621,
-      "step": 124700
+      "learning_rate": 2.771370814071018e-05,
+      "loss": 2.8815,
+      "step": 134500
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7726297603792238e-05,
-      "loss": 2.7572,
-      "step": 124800
+      "learning_rate": 2.7697138406985802e-05,
+      "loss": 2.8914,
+      "step": 134600
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.770845008584656e-05,
-      "loss": 2.759,
-      "step": 124900
+      "learning_rate": 2.768056867326142e-05,
+      "loss": 2.9086,
+      "step": 134700
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.769060256790088e-05,
-      "loss": 2.7554,
-      "step": 125000
+      "learning_rate": 2.7663998939537045e-05,
+      "loss": 2.8938,
+      "step": 134800
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7672755049955202e-05,
-      "loss": 2.7481,
-      "step": 125100
+      "learning_rate": 2.7647429205812663e-05,
+      "loss": 2.9009,
+      "step": 134900
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7654907532009526e-05,
-      "loss": 2.7598,
-      "step": 125200
+      "learning_rate": 2.763085947208828e-05,
+      "loss": 2.9045,
+      "step": 135000
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7637060014063848e-05,
-      "loss": 2.7662,
-      "step": 125300
+      "learning_rate": 2.7614289738363906e-05,
+      "loss": 2.9016,
+      "step": 135100
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.761921249611817e-05,
-      "loss": 2.765,
-      "step": 125400
+      "learning_rate": 2.7597720004639528e-05,
+      "loss": 2.8756,
+      "step": 135200
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.760136497817249e-05,
-      "loss": 2.7548,
-      "step": 125500
+      "learning_rate": 2.7581150270915146e-05,
+      "loss": 2.8986,
+      "step": 135300
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.758351746022681e-05,
-      "loss": 2.7529,
-      "step": 125600
+      "learning_rate": 2.756458053719077e-05,
+      "loss": 2.8812,
+      "step": 135400
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.756566994228113e-05,
-      "loss": 2.7479,
-      "step": 125700
+      "learning_rate": 2.754801080346639e-05,
+      "loss": 2.8877,
+      "step": 135500
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.754782242433545e-05,
-      "loss": 2.7477,
-      "step": 125800
+      "learning_rate": 2.753144106974201e-05,
+      "loss": 2.8851,
+      "step": 135600
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.752997490638977e-05,
-      "loss": 2.745,
-      "step": 125900
+      "learning_rate": 2.7514871336017635e-05,
+      "loss": 2.8811,
+      "step": 135700
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7512127388444093e-05,
-      "loss": 2.7561,
-      "step": 126000
+      "learning_rate": 2.7498301602293253e-05,
+      "loss": 2.8805,
+      "step": 135800
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7494279870498414e-05,
-      "loss": 2.7503,
-      "step": 126100
+      "learning_rate": 2.748173186856887e-05,
+      "loss": 2.8816,
+      "step": 135900
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7476432352552732e-05,
-      "loss": 2.7533,
-      "step": 126200
+      "learning_rate": 2.7465162134844496e-05,
+      "loss": 2.8867,
+      "step": 136000
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7458584834607053e-05,
-      "loss": 2.7483,
-      "step": 126300
+      "learning_rate": 2.7448592401120114e-05,
+      "loss": 2.8897,
+      "step": 136100
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7440737316661374e-05,
-      "loss": 2.7628,
-      "step": 126400
+      "learning_rate": 2.7432022667395736e-05,
+      "loss": 2.887,
+      "step": 136200
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7422889798715696e-05,
-      "loss": 2.7571,
-      "step": 126500
+      "learning_rate": 2.741545293367136e-05,
+      "loss": 2.8932,
+      "step": 136300
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7405042280770017e-05,
-      "loss": 2.7596,
-      "step": 126600
+      "learning_rate": 2.739888319994698e-05,
+      "loss": 2.8946,
+      "step": 136400
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7387194762824338e-05,
-      "loss": 2.7522,
-      "step": 126700
+      "learning_rate": 2.7382313466222597e-05,
+      "loss": 2.8768,
+      "step": 136500
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7369347244878656e-05,
-      "loss": 2.7577,
-      "step": 126800
+      "learning_rate": 2.736574373249822e-05,
+      "loss": 2.8954,
+      "step": 136600
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7351499726932977e-05,
-      "loss": 2.7501,
-      "step": 126900
+      "learning_rate": 2.7349173998773843e-05,
+      "loss": 2.8879,
+      "step": 136700
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.7333652208987298e-05,
-      "loss": 2.7453,
-      "step": 127000
+      "learning_rate": 2.733260426504946e-05,
+      "loss": 2.8942,
+      "step": 136800
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.731580469104162e-05,
-      "loss": 2.7565,
-      "step": 127100
+      "learning_rate": 2.7316034531325086e-05,
+      "loss": 2.8858,
+      "step": 136900
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.729795717309594e-05,
-      "loss": 2.7527,
-      "step": 127200
+      "learning_rate": 2.7299464797600704e-05,
+      "loss": 2.8942,
+      "step": 137000
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.728010965515026e-05,
-      "loss": 2.7426,
-      "step": 127300
+      "learning_rate": 2.7282895063876322e-05,
+      "loss": 2.8879,
+      "step": 137100
     },
     {
       "epoch": 0.45,
-      "learning_rate": 2.726226213720458e-05,
-      "loss": 2.7606,
-      "step": 127400
+      "learning_rate": 2.7266325330151947e-05,
+      "loss": 2.8807,
+      "step": 137200
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.72444146192589e-05,
-      "loss": 2.752,
-      "step": 127500
+      "learning_rate": 2.724975559642757e-05,
+      "loss": 2.8875,
+      "step": 137300
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7226567101313222e-05,
-      "loss": 2.7606,
-      "step": 127600
+      "learning_rate": 2.7233185862703187e-05,
+      "loss": 2.8986,
+      "step": 137400
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7208719583367543e-05,
-      "loss": 2.765,
-      "step": 127700
+      "learning_rate": 2.721661612897881e-05,
+      "loss": 2.8965,
+      "step": 137500
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7190872065421865e-05,
-      "loss": 2.7667,
-      "step": 127800
+      "learning_rate": 2.720004639525443e-05,
+      "loss": 2.8824,
+      "step": 137600
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7173024547476182e-05,
-      "loss": 2.7577,
-      "step": 127900
+      "learning_rate": 2.718347666153005e-05,
+      "loss": 2.8916,
+      "step": 137700
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7155177029530504e-05,
-      "loss": 2.7512,
-      "step": 128000
+      "learning_rate": 2.716690692780567e-05,
+      "loss": 2.8875,
+      "step": 137800
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7137329511584825e-05,
-      "loss": 2.7527,
-      "step": 128100
+      "learning_rate": 2.7150337194081294e-05,
+      "loss": 2.8866,
+      "step": 137900
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7119481993639146e-05,
-      "loss": 2.7469,
-      "step": 128200
+      "learning_rate": 2.7133767460356912e-05,
+      "loss": 2.8806,
+      "step": 138000
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7101634475693467e-05,
-      "loss": 2.7507,
-      "step": 128300
+      "learning_rate": 2.711719772663253e-05,
+      "loss": 2.8949,
+      "step": 138100
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7083786957747785e-05,
-      "loss": 2.7518,
-      "step": 128400
+      "learning_rate": 2.7100627992908155e-05,
+      "loss": 2.8748,
+      "step": 138200
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7065939439802106e-05,
-      "loss": 2.751,
-      "step": 128500
+      "learning_rate": 2.7084058259183776e-05,
+      "loss": 2.8886,
+      "step": 138300
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.7048091921856427e-05,
-      "loss": 2.7627,
-      "step": 128600
+      "learning_rate": 2.7067488525459394e-05,
+      "loss": 2.8927,
+      "step": 138400
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.703024440391075e-05,
-      "loss": 2.7516,
-      "step": 128700
+      "learning_rate": 2.705091879173502e-05,
+      "loss": 2.8764,
+      "step": 138500
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.701239688596507e-05,
-      "loss": 2.7515,
-      "step": 128800
+      "learning_rate": 2.7034349058010637e-05,
+      "loss": 2.8859,
+      "step": 138600
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.699454936801939e-05,
-      "loss": 2.7556,
-      "step": 128900
+      "learning_rate": 2.701777932428626e-05,
+      "loss": 2.8879,
+      "step": 138700
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.697670185007371e-05,
-      "loss": 2.7491,
-      "step": 129000
+      "learning_rate": 2.7001209590561884e-05,
+      "loss": 2.8971,
+      "step": 138800
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.695885433212803e-05,
-      "loss": 2.7524,
-      "step": 129100
+      "learning_rate": 2.6984639856837502e-05,
+      "loss": 2.8799,
+      "step": 138900
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.694100681418235e-05,
-      "loss": 2.7487,
-      "step": 129200
+      "learning_rate": 2.696807012311312e-05,
+      "loss": 2.8932,
+      "step": 139000
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6923159296236673e-05,
-      "loss": 2.763,
-      "step": 129300
+      "learning_rate": 2.6951500389388745e-05,
+      "loss": 2.8941,
+      "step": 139100
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6905311778290994e-05,
-      "loss": 2.739,
-      "step": 129400
+      "learning_rate": 2.6934930655664363e-05,
+      "loss": 2.8961,
+      "step": 139200
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.688746426034531e-05,
-      "loss": 2.7549,
-      "step": 129500
+      "learning_rate": 2.6918360921939984e-05,
+      "loss": 2.8832,
+      "step": 139300
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6869616742399633e-05,
-      "loss": 2.761,
-      "step": 129600
+      "learning_rate": 2.690179118821561e-05,
+      "loss": 2.8821,
+      "step": 139400
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6851769224453954e-05,
-      "loss": 2.7487,
-      "step": 129700
+      "learning_rate": 2.6885221454491227e-05,
+      "loss": 2.8893,
+      "step": 139500
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6833921706508275e-05,
-      "loss": 2.7585,
-      "step": 129800
+      "learning_rate": 2.6868651720766845e-05,
+      "loss": 2.8865,
+      "step": 139600
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6816074188562596e-05,
-      "loss": 2.7546,
-      "step": 129900
+      "learning_rate": 2.685208198704247e-05,
+      "loss": 2.8842,
+      "step": 139700
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6798226670616918e-05,
-      "loss": 2.7487,
-      "step": 130000
+      "learning_rate": 2.683551225331809e-05,
+      "loss": 2.8969,
+      "step": 139800
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6780379152671236e-05,
-      "loss": 2.7442,
-      "step": 130100
+      "learning_rate": 2.681894251959371e-05,
+      "loss": 2.8962,
+      "step": 139900
     },
     {
       "epoch": 0.46,
-      "learning_rate": 2.6762531634725557e-05,
-      "loss": 2.7464,
-      "step": 130200
+      "learning_rate": 2.6802372785869335e-05,
+      "loss": 2.8856,
+      "step": 140000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6785803052144953e-05,
+      "loss": 2.8849,
+      "step": 140100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.676923331842057e-05,
+      "loss": 2.8742,
+      "step": 140200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6752663584696196e-05,
+      "loss": 2.8842,
+      "step": 140300
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6744684116779885e-05,
-      "loss": 2.7618,
-      "step": 130300
+      "learning_rate": 2.6736093850971817e-05,
+      "loss": 2.8989,
+      "step": 140400
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6726836598834203e-05,
-      "loss": 2.7588,
-      "step": 130400
+      "learning_rate": 2.6719524117247435e-05,
+      "loss": 2.885,
+      "step": 140500
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6708989080888524e-05,
-      "loss": 2.7507,
-      "step": 130500
+      "learning_rate": 2.670295438352306e-05,
+      "loss": 2.905,
+      "step": 140600
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6691141562942845e-05,
-      "loss": 2.7527,
-      "step": 130600
+      "learning_rate": 2.6686384649798678e-05,
+      "loss": 2.8828,
+      "step": 140700
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6673294044997166e-05,
-      "loss": 2.7451,
-      "step": 130700
+      "learning_rate": 2.66698149160743e-05,
+      "loss": 2.8825,
+      "step": 140800
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6655446527051487e-05,
-      "loss": 2.747,
-      "step": 130800
+      "learning_rate": 2.6653245182349924e-05,
+      "loss": 2.8992,
+      "step": 140900
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.663759900910581e-05,
-      "loss": 2.7497,
-      "step": 130900
+      "learning_rate": 2.6636675448625542e-05,
+      "loss": 2.8881,
+      "step": 141000
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6619751491160126e-05,
-      "loss": 2.7488,
-      "step": 131000
+      "learning_rate": 2.662010571490116e-05,
+      "loss": 2.9036,
+      "step": 141100
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6601903973214448e-05,
-      "loss": 2.7602,
-      "step": 131100
+      "learning_rate": 2.6603535981176785e-05,
+      "loss": 2.9028,
+      "step": 141200
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.658405645526877e-05,
-      "loss": 2.7607,
-      "step": 131200
+      "learning_rate": 2.6586966247452403e-05,
+      "loss": 2.8867,
+      "step": 141300
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.656620893732309e-05,
-      "loss": 2.755,
-      "step": 131300
+      "learning_rate": 2.6570396513728025e-05,
+      "loss": 2.8796,
+      "step": 141400
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.654836141937741e-05,
-      "loss": 2.7475,
-      "step": 131400
+      "learning_rate": 2.655382678000365e-05,
+      "loss": 2.8807,
+      "step": 141500
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.653051390143173e-05,
-      "loss": 2.748,
-      "step": 131500
+      "learning_rate": 2.6537257046279268e-05,
+      "loss": 2.8988,
+      "step": 141600
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.651266638348605e-05,
-      "loss": 2.7503,
-      "step": 131600
+      "learning_rate": 2.6520687312554886e-05,
+      "loss": 2.8932,
+      "step": 141700
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.649481886554037e-05,
-      "loss": 2.7487,
-      "step": 131700
+      "learning_rate": 2.650411757883051e-05,
+      "loss": 2.8947,
+      "step": 141800
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6476971347594693e-05,
-      "loss": 2.7458,
-      "step": 131800
+      "learning_rate": 2.6487547845106132e-05,
+      "loss": 2.8853,
+      "step": 141900
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6459123829649014e-05,
-      "loss": 2.7588,
-      "step": 131900
+      "learning_rate": 2.647097811138175e-05,
+      "loss": 2.8991,
+      "step": 142000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6454408377657375e-05,
+      "loss": 2.8922,
+      "step": 142100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6437838643932993e-05,
+      "loss": 2.8916,
+      "step": 142200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.642126891020861e-05,
+      "loss": 2.892,
+      "step": 142300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6404699176484236e-05,
+      "loss": 2.8867,
+      "step": 142400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6388129442759858e-05,
+      "loss": 2.8768,
+      "step": 142500
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6441276311703332e-05,
-      "loss": 2.7382,
-      "step": 132000
+      "learning_rate": 2.6371559709035476e-05,
+      "loss": 2.895,
+      "step": 142600
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6423428793757653e-05,
-      "loss": 2.757,
-      "step": 132100
+      "learning_rate": 2.63549899753111e-05,
+      "loss": 2.8872,
+      "step": 142700
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6405581275811974e-05,
-      "loss": 2.762,
-      "step": 132200
+      "learning_rate": 2.633842024158672e-05,
+      "loss": 2.8822,
+      "step": 142800
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6387733757866295e-05,
-      "loss": 2.7397,
-      "step": 132300
+      "learning_rate": 2.632185050786234e-05,
+      "loss": 2.8907,
+      "step": 142900
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6369886239920617e-05,
-      "loss": 2.7499,
-      "step": 132400
+      "learning_rate": 2.6305280774137958e-05,
+      "loss": 2.8897,
+      "step": 143000
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6352038721974938e-05,
-      "loss": 2.7543,
-      "step": 132500
+      "learning_rate": 2.6288711040413583e-05,
+      "loss": 2.8906,
+      "step": 143100
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6334191204029256e-05,
-      "loss": 2.7473,
-      "step": 132600
+      "learning_rate": 2.62721413066892e-05,
+      "loss": 2.8884,
+      "step": 143200
     },
     {
       "epoch": 0.47,
-      "learning_rate": 2.6316343686083577e-05,
-      "loss": 2.7637,
-      "step": 132700
+      "learning_rate": 2.6255571572964823e-05,
+      "loss": 2.892,
+      "step": 143300
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.6298496168137898e-05,
-      "loss": 2.7512,
-      "step": 132800
+      "epoch": 0.48,
+      "learning_rate": 2.6239001839240444e-05,
+      "loss": 2.8883,
+      "step": 143400
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.628064865019222e-05,
-      "loss": 2.7512,
-      "step": 132900
+      "epoch": 0.48,
+      "learning_rate": 2.6222432105516066e-05,
+      "loss": 2.8872,
+      "step": 143500
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.626280113224654e-05,
-      "loss": 2.7439,
-      "step": 133000
+      "epoch": 0.48,
+      "learning_rate": 2.6205862371791684e-05,
+      "loss": 2.8916,
+      "step": 143600
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.624495361430086e-05,
-      "loss": 2.7528,
-      "step": 133100
+      "learning_rate": 2.618929263806731e-05,
+      "loss": 2.8746,
+      "step": 143700
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.622710609635518e-05,
-      "loss": 2.7516,
-      "step": 133200
+      "learning_rate": 2.6172722904342927e-05,
+      "loss": 2.8836,
+      "step": 143800
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.62092585784095e-05,
-      "loss": 2.7441,
-      "step": 133300
+      "learning_rate": 2.6156153170618548e-05,
+      "loss": 2.8954,
+      "step": 143900
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6191411060463822e-05,
-      "loss": 2.7522,
-      "step": 133400
+      "learning_rate": 2.6139583436894173e-05,
+      "loss": 2.8952,
+      "step": 144000
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6173563542518143e-05,
-      "loss": 2.7407,
-      "step": 133500
+      "learning_rate": 2.612301370316979e-05,
+      "loss": 2.8896,
+      "step": 144100
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6155716024572465e-05,
-      "loss": 2.7493,
-      "step": 133600
+      "learning_rate": 2.610644396944541e-05,
+      "loss": 2.8762,
+      "step": 144200
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6137868506626782e-05,
-      "loss": 2.738,
-      "step": 133700
+      "learning_rate": 2.6089874235721034e-05,
+      "loss": 2.8893,
+      "step": 144300
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6120020988681104e-05,
-      "loss": 2.7546,
-      "step": 133800
+      "learning_rate": 2.6073304501996655e-05,
+      "loss": 2.8845,
+      "step": 144400
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6102173470735425e-05,
-      "loss": 2.7533,
-      "step": 133900
+      "learning_rate": 2.6056734768272273e-05,
+      "loss": 2.8777,
+      "step": 144500
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6084325952789746e-05,
-      "loss": 2.7519,
-      "step": 134000
+      "learning_rate": 2.60401650345479e-05,
+      "loss": 2.8806,
+      "step": 144600
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6066478434844067e-05,
-      "loss": 2.7701,
-      "step": 134100
+      "learning_rate": 2.6023595300823516e-05,
+      "loss": 2.8974,
+      "step": 144700
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6048630916898385e-05,
-      "loss": 2.7568,
-      "step": 134200
+      "learning_rate": 2.6007025567099134e-05,
+      "loss": 2.898,
+      "step": 144800
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6030783398952706e-05,
-      "loss": 2.7504,
-      "step": 134300
+      "learning_rate": 2.599045583337476e-05,
+      "loss": 2.8857,
+      "step": 144900
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.6012935881007027e-05,
-      "loss": 2.7449,
-      "step": 134400
+      "learning_rate": 2.597388609965038e-05,
+      "loss": 2.8771,
+      "step": 145000
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.599508836306135e-05,
-      "loss": 2.7515,
-      "step": 134500
+      "learning_rate": 2.5957316365926e-05,
+      "loss": 2.8773,
+      "step": 145100
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.597724084511567e-05,
-      "loss": 2.751,
-      "step": 134600
+      "learning_rate": 2.5940746632201624e-05,
+      "loss": 2.885,
+      "step": 145200
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.595939332716999e-05,
-      "loss": 2.7471,
-      "step": 134700
+      "learning_rate": 2.5924176898477242e-05,
+      "loss": 2.8782,
+      "step": 145300
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.594154580922431e-05,
-      "loss": 2.7567,
-      "step": 134800
+      "learning_rate": 2.5907607164752863e-05,
+      "loss": 2.8866,
+      "step": 145400
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.592369829127863e-05,
-      "loss": 2.7525,
-      "step": 134900
+      "learning_rate": 2.5891037431028485e-05,
+      "loss": 2.8881,
+      "step": 145500
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.590585077333295e-05,
-      "loss": 2.7495,
-      "step": 135000
+      "learning_rate": 2.5874467697304106e-05,
+      "loss": 2.8791,
+      "step": 145600
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.5888003255387273e-05,
-      "loss": 2.7366,
-      "step": 135100
+      "learning_rate": 2.5857897963579724e-05,
+      "loss": 2.8828,
+      "step": 145700
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.5870155737441594e-05,
-      "loss": 2.7473,
-      "step": 135200
+      "learning_rate": 2.584132822985535e-05,
+      "loss": 2.9043,
+      "step": 145800
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.585230821949591e-05,
-      "loss": 2.7443,
-      "step": 135300
+      "learning_rate": 2.5824758496130967e-05,
+      "loss": 2.8929,
+      "step": 145900
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.583446070155024e-05,
-      "loss": 2.7467,
-      "step": 135400
+      "learning_rate": 2.580818876240659e-05,
+      "loss": 2.8861,
+      "step": 146000
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.581661318360456e-05,
-      "loss": 2.738,
-      "step": 135500
+      "learning_rate": 2.5791619028682214e-05,
+      "loss": 2.8801,
+      "step": 146100
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.5798765665658882e-05,
-      "loss": 2.7387,
-      "step": 135600
+      "learning_rate": 2.577504929495783e-05,
+      "loss": 2.8804,
+      "step": 146200
     },
     {
       "epoch": 0.48,
-      "learning_rate": 2.57809181477132e-05,
-      "loss": 2.7489,
-      "step": 135700
+      "learning_rate": 2.575847956123345e-05,
+      "loss": 2.882,
+      "step": 146300
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 2.576307062976752e-05,
-      "loss": 2.7453,
-      "step": 135800
+      "epoch": 0.49,
+      "learning_rate": 2.5741909827509075e-05,
+      "loss": 2.879,
+      "step": 146400
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5745223111821842e-05,
-      "loss": 2.7451,
-      "step": 135900
+      "learning_rate": 2.5725340093784696e-05,
+      "loss": 2.8754,
+      "step": 146500
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5727375593876163e-05,
-      "loss": 2.7548,
-      "step": 136000
+      "learning_rate": 2.5708770360060314e-05,
+      "loss": 2.8907,
+      "step": 146600
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5709528075930485e-05,
-      "loss": 2.7495,
-      "step": 136100
+      "learning_rate": 2.569220062633594e-05,
+      "loss": 2.8898,
+      "step": 146700
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5691680557984803e-05,
-      "loss": 2.7477,
-      "step": 136200
+      "learning_rate": 2.5675630892611557e-05,
+      "loss": 2.8777,
+      "step": 146800
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5673833040039124e-05,
-      "loss": 2.7342,
-      "step": 136300
+      "learning_rate": 2.5659061158887175e-05,
+      "loss": 2.8793,
+      "step": 146900
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5655985522093445e-05,
-      "loss": 2.747,
-      "step": 136400
+      "learning_rate": 2.56424914251628e-05,
+      "loss": 2.8791,
+      "step": 147000
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5638138004147766e-05,
-      "loss": 2.7491,
-      "step": 136500
+      "learning_rate": 2.562592169143842e-05,
+      "loss": 2.8859,
+      "step": 147100
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5620290486202087e-05,
-      "loss": 2.7354,
-      "step": 136600
+      "learning_rate": 2.560935195771404e-05,
+      "loss": 2.88,
+      "step": 147200
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.560244296825641e-05,
-      "loss": 2.7453,
-      "step": 136700
+      "learning_rate": 2.5592782223989664e-05,
+      "loss": 2.8931,
+      "step": 147300
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5584595450310726e-05,
-      "loss": 2.7476,
-      "step": 136800
+      "learning_rate": 2.5576212490265282e-05,
+      "loss": 2.8722,
+      "step": 147400
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5566747932365048e-05,
-      "loss": 2.7495,
-      "step": 136900
+      "learning_rate": 2.5559642756540904e-05,
+      "loss": 2.8907,
+      "step": 147500
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.554890041441937e-05,
-      "loss": 2.7475,
-      "step": 137000
+      "learning_rate": 2.554307302281653e-05,
+      "loss": 2.8774,
+      "step": 147600
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.553105289647369e-05,
-      "loss": 2.7538,
-      "step": 137100
+      "learning_rate": 2.5526503289092147e-05,
+      "loss": 2.8912,
+      "step": 147700
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.551320537852801e-05,
-      "loss": 2.7488,
-      "step": 137200
+      "learning_rate": 2.5509933555367765e-05,
+      "loss": 2.8937,
+      "step": 147800
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.549535786058233e-05,
-      "loss": 2.7532,
-      "step": 137300
+      "learning_rate": 2.549336382164339e-05,
+      "loss": 2.8644,
+      "step": 147900
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.547751034263665e-05,
-      "loss": 2.7602,
-      "step": 137400
+      "learning_rate": 2.5476794087919008e-05,
+      "loss": 2.8743,
+      "step": 148000
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.545966282469097e-05,
-      "loss": 2.7702,
-      "step": 137500
+      "learning_rate": 2.546022435419463e-05,
+      "loss": 2.8732,
+      "step": 148100
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5441815306745293e-05,
-      "loss": 2.7445,
-      "step": 137600
+      "learning_rate": 2.5443654620470247e-05,
+      "loss": 2.8912,
+      "step": 148200
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5423967788799614e-05,
-      "loss": 2.7486,
-      "step": 137700
+      "learning_rate": 2.5427084886745872e-05,
+      "loss": 2.886,
+      "step": 148300
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5406120270853935e-05,
-      "loss": 2.7399,
-      "step": 137800
+      "learning_rate": 2.541051515302149e-05,
+      "loss": 2.8745,
+      "step": 148400
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5388272752908253e-05,
-      "loss": 2.7414,
-      "step": 137900
+      "learning_rate": 2.5393945419297112e-05,
+      "loss": 2.8838,
+      "step": 148500
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5370425234962574e-05,
-      "loss": 2.7525,
-      "step": 138000
+      "learning_rate": 2.5377375685572737e-05,
+      "loss": 2.8772,
+      "step": 148600
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5352577717016895e-05,
-      "loss": 2.7458,
-      "step": 138100
+      "learning_rate": 2.5360805951848355e-05,
+      "loss": 2.8788,
+      "step": 148700
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5334730199071217e-05,
-      "loss": 2.752,
-      "step": 138200
+      "learning_rate": 2.5344236218123973e-05,
+      "loss": 2.8653,
+      "step": 148800
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5316882681125538e-05,
-      "loss": 2.7483,
-      "step": 138300
+      "learning_rate": 2.5327666484399598e-05,
+      "loss": 2.8815,
+      "step": 148900
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5299035163179856e-05,
-      "loss": 2.74,
-      "step": 138400
+      "learning_rate": 2.5311096750675216e-05,
+      "loss": 2.8838,
+      "step": 149000
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5281187645234177e-05,
-      "loss": 2.7632,
-      "step": 138500
+      "learning_rate": 2.5294527016950837e-05,
+      "loss": 2.876,
+      "step": 149100
     },
     {
       "epoch": 0.49,
-      "learning_rate": 2.5263340127288498e-05,
-      "loss": 2.7447,
-      "step": 138600
+      "learning_rate": 2.5277957283226462e-05,
+      "loss": 2.8888,
+      "step": 149200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.526138754950208e-05,
+      "loss": 2.8827,
+      "step": 149300
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.524549260934282e-05,
-      "loss": 2.7355,
-      "step": 138700
+      "learning_rate": 2.5244817815777698e-05,
+      "loss": 2.8614,
+      "step": 149400
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.522764509139714e-05,
-      "loss": 2.7461,
-      "step": 138800
+      "learning_rate": 2.5228248082053323e-05,
+      "loss": 2.8841,
+      "step": 149500
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5209797573451462e-05,
-      "loss": 2.7478,
-      "step": 138900
+      "learning_rate": 2.5211678348328945e-05,
+      "loss": 2.8911,
+      "step": 149600
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.519195005550578e-05,
-      "loss": 2.7389,
-      "step": 139000
+      "learning_rate": 2.5195108614604563e-05,
+      "loss": 2.8801,
+      "step": 149700
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.51741025375601e-05,
-      "loss": 2.7421,
-      "step": 139100
+      "learning_rate": 2.5178538880880188e-05,
+      "loss": 2.8791,
+      "step": 149800
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5156255019614422e-05,
-      "loss": 2.7595,
-      "step": 139200
+      "learning_rate": 2.5161969147155806e-05,
+      "loss": 2.8855,
+      "step": 149900
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5138407501668743e-05,
-      "loss": 2.7452,
-      "step": 139300
+      "learning_rate": 2.5145399413431424e-05,
+      "loss": 2.8709,
+      "step": 150000
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5120559983723064e-05,
-      "loss": 2.746,
-      "step": 139400
+      "learning_rate": 2.512882967970705e-05,
+      "loss": 2.8825,
+      "step": 150100
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5102712465777382e-05,
-      "loss": 2.7434,
-      "step": 139500
+      "learning_rate": 2.511225994598267e-05,
+      "loss": 2.8924,
+      "step": 150200
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5084864947831704e-05,
-      "loss": 2.7466,
-      "step": 139600
+      "learning_rate": 2.5095690212258288e-05,
+      "loss": 2.8547,
+      "step": 150300
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5067017429886025e-05,
-      "loss": 2.7494,
-      "step": 139700
+      "learning_rate": 2.5079120478533913e-05,
+      "loss": 2.8808,
+      "step": 150400
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5049169911940346e-05,
-      "loss": 2.7442,
-      "step": 139800
+      "learning_rate": 2.506255074480953e-05,
+      "loss": 2.8742,
+      "step": 150500
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.5031322393994667e-05,
-      "loss": 2.736,
-      "step": 139900
+      "learning_rate": 2.5045981011085152e-05,
+      "loss": 2.8711,
+      "step": 150600
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.501347487604899e-05,
-      "loss": 2.7473,
-      "step": 140000
+      "learning_rate": 2.5029411277360777e-05,
+      "loss": 2.8726,
+      "step": 150700
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.499562735810331e-05,
-      "loss": 2.7432,
-      "step": 140100
+      "learning_rate": 2.5012841543636395e-05,
+      "loss": 2.8841,
+      "step": 150800
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.497777984015763e-05,
-      "loss": 2.7406,
-      "step": 140200
+      "learning_rate": 2.4996271809912017e-05,
+      "loss": 2.8789,
+      "step": 150900
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4959932322211952e-05,
-      "loss": 2.7418,
-      "step": 140300
+      "learning_rate": 2.4979702076187635e-05,
+      "loss": 2.8816,
+      "step": 151000
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4942084804266273e-05,
-      "loss": 2.7518,
-      "step": 140400
+      "learning_rate": 2.4963132342463256e-05,
+      "loss": 2.8908,
+      "step": 151100
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.492423728632059e-05,
-      "loss": 2.7358,
-      "step": 140500
+      "learning_rate": 2.494656260873888e-05,
+      "loss": 2.8838,
+      "step": 151200
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4906389768374912e-05,
-      "loss": 2.7471,
-      "step": 140600
+      "learning_rate": 2.49299928750145e-05,
+      "loss": 2.8876,
+      "step": 151300
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4888542250429233e-05,
-      "loss": 2.7565,
-      "step": 140700
+      "learning_rate": 2.491342314129012e-05,
+      "loss": 2.8947,
+      "step": 151400
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4870694732483555e-05,
-      "loss": 2.7375,
-      "step": 140800
+      "learning_rate": 2.4896853407565742e-05,
+      "loss": 2.8873,
+      "step": 151500
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4852847214537876e-05,
-      "loss": 2.7518,
-      "step": 140900
+      "learning_rate": 2.488028367384136e-05,
+      "loss": 2.8807,
+      "step": 151600
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4834999696592197e-05,
-      "loss": 2.7417,
-      "step": 141000
+      "learning_rate": 2.4863713940116985e-05,
+      "loss": 2.8885,
+      "step": 151700
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4817152178646515e-05,
-      "loss": 2.7378,
-      "step": 141100
+      "learning_rate": 2.4847144206392607e-05,
+      "loss": 2.8839,
+      "step": 151800
     },
     {
       "epoch": 0.5,
-      "learning_rate": 2.4799304660700836e-05,
-      "loss": 2.7404,
-      "step": 141200
+      "learning_rate": 2.4830574472668225e-05,
+      "loss": 2.8794,
+      "step": 151900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4814004738943846e-05,
+      "loss": 2.881,
+      "step": 152000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4797435005219464e-05,
+      "loss": 2.8763,
+      "step": 152100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.478086527149509e-05,
+      "loss": 2.8905,
+      "step": 152200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.476429553777071e-05,
+      "loss": 2.8765,
+      "step": 152300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.474772580404633e-05,
+      "loss": 2.8894,
+      "step": 152400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.473115607032195e-05,
+      "loss": 2.8771,
+      "step": 152500
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.4781457142755157e-05,
-      "loss": 2.7438,
-      "step": 141300
+      "epoch": 0.51,
+      "learning_rate": 2.471458633659757e-05,
+      "loss": 2.8776,
+      "step": 152600
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.476360962480948e-05,
-      "loss": 2.7558,
-      "step": 141400
+      "epoch": 0.51,
+      "learning_rate": 2.4698016602873193e-05,
+      "loss": 2.876,
+      "step": 152700
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.47457621068638e-05,
-      "loss": 2.7436,
-      "step": 141500
+      "learning_rate": 2.4681446869148815e-05,
+      "loss": 2.8804,
+      "step": 152800
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4727914588918118e-05,
-      "loss": 2.7479,
-      "step": 141600
+      "learning_rate": 2.4664877135424436e-05,
+      "loss": 2.889,
+      "step": 152900
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.471006707097244e-05,
-      "loss": 2.7583,
-      "step": 141700
+      "learning_rate": 2.4648307401700054e-05,
+      "loss": 2.886,
+      "step": 153000
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4692219553026763e-05,
-      "loss": 2.7381,
-      "step": 141800
+      "learning_rate": 2.4631737667975676e-05,
+      "loss": 2.8929,
+      "step": 153100
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4674372035081085e-05,
-      "loss": 2.7397,
-      "step": 141900
+      "learning_rate": 2.4615167934251297e-05,
+      "loss": 2.8812,
+      "step": 153200
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4656524517135406e-05,
-      "loss": 2.744,
-      "step": 142000
+      "learning_rate": 2.459859820052692e-05,
+      "loss": 2.8745,
+      "step": 153300
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4638676999189724e-05,
-      "loss": 2.7566,
-      "step": 142100
+      "learning_rate": 2.458202846680254e-05,
+      "loss": 2.88,
+      "step": 153400
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4620829481244045e-05,
-      "loss": 2.7493,
-      "step": 142200
+      "learning_rate": 2.456545873307816e-05,
+      "loss": 2.8788,
+      "step": 153500
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4602981963298366e-05,
-      "loss": 2.7335,
-      "step": 142300
+      "learning_rate": 2.454888899935378e-05,
+      "loss": 2.8796,
+      "step": 153600
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4585134445352687e-05,
-      "loss": 2.7386,
-      "step": 142400
+      "learning_rate": 2.45323192656294e-05,
+      "loss": 2.882,
+      "step": 153700
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.456728692740701e-05,
-      "loss": 2.7502,
-      "step": 142500
+      "learning_rate": 2.4515749531905026e-05,
+      "loss": 2.8739,
+      "step": 153800
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4549439409461326e-05,
-      "loss": 2.7461,
-      "step": 142600
+      "learning_rate": 2.4499179798180644e-05,
+      "loss": 2.8734,
+      "step": 153900
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4531591891515648e-05,
-      "loss": 2.7352,
-      "step": 142700
+      "learning_rate": 2.4482610064456265e-05,
+      "loss": 2.8791,
+      "step": 154000
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.451374437356997e-05,
-      "loss": 2.7417,
-      "step": 142800
+      "learning_rate": 2.4466040330731887e-05,
+      "loss": 2.8756,
+      "step": 154100
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.449589685562429e-05,
-      "loss": 2.7435,
-      "step": 142900
+      "learning_rate": 2.4449470597007505e-05,
+      "loss": 2.8882,
+      "step": 154200
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.447804933767861e-05,
-      "loss": 2.748,
-      "step": 143000
+      "learning_rate": 2.443290086328313e-05,
+      "loss": 2.8941,
+      "step": 154300
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.446020181973293e-05,
-      "loss": 2.7458,
-      "step": 143100
+      "learning_rate": 2.441633112955875e-05,
+      "loss": 2.8856,
+      "step": 154400
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.444235430178725e-05,
-      "loss": 2.7492,
-      "step": 143200
+      "learning_rate": 2.439976139583437e-05,
+      "loss": 2.8739,
+      "step": 154500
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.442450678384157e-05,
-      "loss": 2.7469,
-      "step": 143300
+      "learning_rate": 2.438319166210999e-05,
+      "loss": 2.8941,
+      "step": 154600
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4406659265895893e-05,
-      "loss": 2.7519,
-      "step": 143400
+      "learning_rate": 2.4366621928385612e-05,
+      "loss": 2.8838,
+      "step": 154700
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4388811747950214e-05,
-      "loss": 2.7411,
-      "step": 143500
+      "learning_rate": 2.4350052194661234e-05,
+      "loss": 2.8642,
+      "step": 154800
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4370964230004535e-05,
-      "loss": 2.7353,
-      "step": 143600
+      "learning_rate": 2.4333482460936855e-05,
+      "loss": 2.8831,
+      "step": 154900
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4353116712058853e-05,
-      "loss": 2.7478,
-      "step": 143700
+      "learning_rate": 2.4316912727212473e-05,
+      "loss": 2.8704,
+      "step": 155000
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4335269194113174e-05,
-      "loss": 2.7418,
-      "step": 143800
+      "learning_rate": 2.4300342993488095e-05,
+      "loss": 2.8804,
+      "step": 155100
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4317421676167495e-05,
-      "loss": 2.7269,
-      "step": 143900
+      "learning_rate": 2.4283773259763716e-05,
+      "loss": 2.8601,
+      "step": 155200
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4299574158221817e-05,
-      "loss": 2.7446,
-      "step": 144000
+      "learning_rate": 2.4267203526039338e-05,
+      "loss": 2.8754,
+      "step": 155300
     },
     {
       "epoch": 0.51,
-      "learning_rate": 2.4281726640276138e-05,
-      "loss": 2.7508,
-      "step": 144100
+      "learning_rate": 2.425063379231496e-05,
+      "loss": 2.8693,
+      "step": 155400
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 2.4263879122330456e-05,
-      "loss": 2.7431,
-      "step": 144200
+      "epoch": 0.52,
+      "learning_rate": 2.423406405859058e-05,
+      "loss": 2.8839,
+      "step": 155500
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.424603160438478e-05,
-      "loss": 2.7478,
-      "step": 144300
+      "learning_rate": 2.42174943248662e-05,
+      "loss": 2.8934,
+      "step": 155600
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.42281840864391e-05,
-      "loss": 2.7388,
-      "step": 144400
+      "learning_rate": 2.420092459114182e-05,
+      "loss": 2.878,
+      "step": 155700
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4210336568493423e-05,
-      "loss": 2.7506,
-      "step": 144500
+      "learning_rate": 2.418435485741744e-05,
+      "loss": 2.8855,
+      "step": 155800
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4192489050547744e-05,
-      "loss": 2.7405,
-      "step": 144600
+      "learning_rate": 2.4167785123693063e-05,
+      "loss": 2.8888,
+      "step": 155900
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4174641532602062e-05,
-      "loss": 2.7351,
-      "step": 144700
+      "learning_rate": 2.4151215389968685e-05,
+      "loss": 2.8654,
+      "step": 156000
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4156794014656383e-05,
-      "loss": 2.7508,
-      "step": 144800
+      "learning_rate": 2.4134645656244306e-05,
+      "loss": 2.874,
+      "step": 156100
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4138946496710704e-05,
-      "loss": 2.7595,
-      "step": 144900
+      "learning_rate": 2.4118075922519924e-05,
+      "loss": 2.8827,
+      "step": 156200
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4121098978765025e-05,
-      "loss": 2.7456,
-      "step": 145000
+      "learning_rate": 2.410150618879555e-05,
+      "loss": 2.8718,
+      "step": 156300
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4103251460819347e-05,
-      "loss": 2.7487,
-      "step": 145100
+      "learning_rate": 2.408493645507117e-05,
+      "loss": 2.874,
+      "step": 156400
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4085403942873664e-05,
-      "loss": 2.7473,
-      "step": 145200
+      "learning_rate": 2.406836672134679e-05,
+      "loss": 2.8826,
+      "step": 156500
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4067556424927986e-05,
-      "loss": 2.7426,
-      "step": 145300
+      "learning_rate": 2.405179698762241e-05,
+      "loss": 2.8972,
+      "step": 156600
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4049708906982307e-05,
-      "loss": 2.7394,
-      "step": 145400
+      "learning_rate": 2.403522725389803e-05,
+      "loss": 2.873,
+      "step": 156700
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.4031861389036628e-05,
-      "loss": 2.7362,
-      "step": 145500
+      "learning_rate": 2.4018657520173653e-05,
+      "loss": 2.875,
+      "step": 156800
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.401401387109095e-05,
-      "loss": 2.7418,
-      "step": 145600
+      "learning_rate": 2.4002087786449274e-05,
+      "loss": 2.8828,
+      "step": 156900
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.399616635314527e-05,
-      "loss": 2.7537,
-      "step": 145700
+      "learning_rate": 2.3985518052724896e-05,
+      "loss": 2.8807,
+      "step": 157000
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.397831883519959e-05,
-      "loss": 2.7451,
-      "step": 145800
+      "learning_rate": 2.3968948319000514e-05,
+      "loss": 2.8767,
+      "step": 157100
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.396047131725391e-05,
-      "loss": 2.7504,
-      "step": 145900
+      "learning_rate": 2.3952378585276135e-05,
+      "loss": 2.882,
+      "step": 157200
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.394262379930823e-05,
-      "loss": 2.7472,
-      "step": 146000
+      "learning_rate": 2.3935808851551757e-05,
+      "loss": 2.8868,
+      "step": 157300
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3924776281362552e-05,
-      "loss": 2.7374,
-      "step": 146100
+      "learning_rate": 2.391923911782738e-05,
+      "loss": 2.8646,
+      "step": 157400
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3906928763416873e-05,
-      "loss": 2.7421,
-      "step": 146200
+      "learning_rate": 2.3902669384103e-05,
+      "loss": 2.8824,
+      "step": 157500
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.388908124547119e-05,
-      "loss": 2.745,
-      "step": 146300
+      "learning_rate": 2.3886099650378618e-05,
+      "loss": 2.8692,
+      "step": 157600
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3871233727525512e-05,
-      "loss": 2.7375,
-      "step": 146400
+      "learning_rate": 2.386952991665424e-05,
+      "loss": 2.8803,
+      "step": 157700
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3853386209579833e-05,
-      "loss": 2.7346,
-      "step": 146500
+      "learning_rate": 2.385296018292986e-05,
+      "loss": 2.8792,
+      "step": 157800
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3835538691634155e-05,
-      "loss": 2.7361,
-      "step": 146600
+      "learning_rate": 2.3836390449205482e-05,
+      "loss": 2.8663,
+      "step": 157900
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3817691173688476e-05,
-      "loss": 2.7455,
-      "step": 146700
+      "learning_rate": 2.3819820715481104e-05,
+      "loss": 2.8776,
+      "step": 158000
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.3799843655742797e-05,
-      "loss": 2.7492,
-      "step": 146800
+      "learning_rate": 2.3803250981756725e-05,
+      "loss": 2.8769,
+      "step": 158100
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.378199613779712e-05,
-      "loss": 2.7403,
-      "step": 146900
+      "learning_rate": 2.3786681248032343e-05,
+      "loss": 2.8845,
+      "step": 158200
     },
     {
       "epoch": 0.52,
-      "learning_rate": 2.376414861985144e-05,
-      "loss": 2.739,
-      "step": 147000
+      "learning_rate": 2.3770111514307965e-05,
+      "loss": 2.8731,
+      "step": 158300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.375354178058359e-05,
+      "loss": 2.8754,
+      "step": 158400
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.374630110190576e-05,
-      "loss": 2.7435,
-      "step": 147100
+      "learning_rate": 2.3736972046859208e-05,
+      "loss": 2.8688,
+      "step": 158500
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3728453583960082e-05,
-      "loss": 2.7487,
-      "step": 147200
+      "learning_rate": 2.372040231313483e-05,
+      "loss": 2.8834,
+      "step": 158600
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.37106060660144e-05,
-      "loss": 2.7402,
-      "step": 147300
+      "learning_rate": 2.370383257941045e-05,
+      "loss": 2.8803,
+      "step": 158700
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.369275854806872e-05,
-      "loss": 2.7438,
-      "step": 147400
+      "learning_rate": 2.368726284568607e-05,
+      "loss": 2.8574,
+      "step": 158800
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3674911030123042e-05,
-      "loss": 2.7439,
-      "step": 147500
+      "learning_rate": 2.3670693111961694e-05,
+      "loss": 2.8729,
+      "step": 158900
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3657063512177363e-05,
-      "loss": 2.7467,
-      "step": 147600
+      "learning_rate": 2.3654123378237315e-05,
+      "loss": 2.8787,
+      "step": 159000
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3639215994231685e-05,
-      "loss": 2.7309,
-      "step": 147700
+      "learning_rate": 2.3637553644512933e-05,
+      "loss": 2.8722,
+      "step": 159100
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3621368476286006e-05,
-      "loss": 2.7259,
-      "step": 147800
+      "learning_rate": 2.3620983910788555e-05,
+      "loss": 2.8869,
+      "step": 159200
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3603520958340324e-05,
-      "loss": 2.7411,
-      "step": 147900
+      "learning_rate": 2.3604414177064176e-05,
+      "loss": 2.8852,
+      "step": 159300
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3585673440394645e-05,
-      "loss": 2.731,
-      "step": 148000
+      "learning_rate": 2.3587844443339798e-05,
+      "loss": 2.8793,
+      "step": 159400
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3567825922448966e-05,
-      "loss": 2.7355,
-      "step": 148100
+      "learning_rate": 2.357127470961542e-05,
+      "loss": 2.8703,
+      "step": 159500
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3549978404503287e-05,
-      "loss": 2.7483,
-      "step": 148200
+      "learning_rate": 2.355470497589104e-05,
+      "loss": 2.8729,
+      "step": 159600
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.353213088655761e-05,
-      "loss": 2.7381,
-      "step": 148300
+      "learning_rate": 2.353813524216666e-05,
+      "loss": 2.8683,
+      "step": 159700
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3514283368611926e-05,
-      "loss": 2.7356,
-      "step": 148400
+      "learning_rate": 2.352156550844228e-05,
+      "loss": 2.8763,
+      "step": 159800
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3496435850666248e-05,
-      "loss": 2.7512,
-      "step": 148500
+      "learning_rate": 2.35049957747179e-05,
+      "loss": 2.8692,
+      "step": 159900
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.347858833272057e-05,
-      "loss": 2.7407,
-      "step": 148600
+      "learning_rate": 2.3488426040993523e-05,
+      "loss": 2.8703,
+      "step": 160000
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.346074081477489e-05,
-      "loss": 2.7338,
-      "step": 148700
+      "learning_rate": 2.3471856307269144e-05,
+      "loss": 2.8631,
+      "step": 160100
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.344289329682921e-05,
-      "loss": 2.7281,
-      "step": 148800
+      "learning_rate": 2.3455286573544762e-05,
+      "loss": 2.8788,
+      "step": 160200
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3425045778883532e-05,
-      "loss": 2.7326,
-      "step": 148900
+      "learning_rate": 2.3438716839820384e-05,
+      "loss": 2.8734,
+      "step": 160300
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.340719826093785e-05,
-      "loss": 2.7227,
-      "step": 149000
+      "learning_rate": 2.3422147106096005e-05,
+      "loss": 2.8666,
+      "step": 160400
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.338935074299217e-05,
-      "loss": 2.7437,
-      "step": 149100
+      "learning_rate": 2.3405577372371627e-05,
+      "loss": 2.8745,
+      "step": 160500
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3371503225046493e-05,
-      "loss": 2.739,
-      "step": 149200
+      "learning_rate": 2.338900763864725e-05,
+      "loss": 2.872,
+      "step": 160600
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3353655707100814e-05,
-      "loss": 2.7323,
-      "step": 149300
+      "learning_rate": 2.337243790492287e-05,
+      "loss": 2.8738,
+      "step": 160700
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3335808189155135e-05,
-      "loss": 2.7362,
-      "step": 149400
+      "learning_rate": 2.3355868171198488e-05,
+      "loss": 2.8777,
+      "step": 160800
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3317960671209456e-05,
-      "loss": 2.7386,
-      "step": 149500
+      "learning_rate": 2.333929843747411e-05,
+      "loss": 2.8737,
+      "step": 160900
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.3300113153263778e-05,
-      "loss": 2.7508,
-      "step": 149600
+      "learning_rate": 2.3322728703749734e-05,
+      "loss": 2.876,
+      "step": 161000
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.32822656353181e-05,
-      "loss": 2.7415,
-      "step": 149700
+      "learning_rate": 2.3306158970025352e-05,
+      "loss": 2.886,
+      "step": 161100
     },
     {
       "epoch": 0.53,
-      "learning_rate": 2.326441811737242e-05,
-      "loss": 2.7551,
-      "step": 149800
+      "learning_rate": 2.3289589236300974e-05,
+      "loss": 2.876,
+      "step": 161200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3273019502576595e-05,
+      "loss": 2.8769,
+      "step": 161300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3256449768852213e-05,
+      "loss": 2.8749,
+      "step": 161400
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.324657059942674e-05,
-      "loss": 2.7312,
-      "step": 149900
+      "learning_rate": 2.3239880035127838e-05,
+      "loss": 2.8808,
+      "step": 161500
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.322872308148106e-05,
-      "loss": 2.7526,
-      "step": 150000
+      "learning_rate": 2.322331030140346e-05,
+      "loss": 2.8681,
+      "step": 161600
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.321087556353538e-05,
-      "loss": 2.7379,
-      "step": 150100
+      "learning_rate": 2.3206740567679078e-05,
+      "loss": 2.8682,
+      "step": 161700
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.31930280455897e-05,
-      "loss": 2.7416,
-      "step": 150200
+      "learning_rate": 2.31901708339547e-05,
+      "loss": 2.8795,
+      "step": 161800
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3175180527644023e-05,
-      "loss": 2.7331,
-      "step": 150300
+      "learning_rate": 2.317360110023032e-05,
+      "loss": 2.8793,
+      "step": 161900
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3157333009698344e-05,
-      "loss": 2.7272,
-      "step": 150400
+      "learning_rate": 2.3157031366505942e-05,
+      "loss": 2.863,
+      "step": 162000
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3139485491752662e-05,
-      "loss": 2.7433,
-      "step": 150500
+      "learning_rate": 2.3140461632781564e-05,
+      "loss": 2.8749,
+      "step": 162100
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3121637973806983e-05,
-      "loss": 2.7358,
-      "step": 150600
+      "learning_rate": 2.3123891899057185e-05,
+      "loss": 2.8735,
+      "step": 162200
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3103790455861304e-05,
-      "loss": 2.7218,
-      "step": 150700
+      "learning_rate": 2.3107322165332803e-05,
+      "loss": 2.8774,
+      "step": 162300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3090752431608425e-05,
+      "loss": 2.8796,
+      "step": 162400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3074182697884046e-05,
+      "loss": 2.872,
+      "step": 162500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3057612964159668e-05,
+      "loss": 2.8836,
+      "step": 162600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.304104323043529e-05,
+      "loss": 2.8775,
+      "step": 162700
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3085942937915625e-05,
-      "loss": 2.73,
-      "step": 150800
+      "learning_rate": 2.3024473496710907e-05,
+      "loss": 2.8755,
+      "step": 162800
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3068095419969947e-05,
-      "loss": 2.734,
-      "step": 150900
+      "learning_rate": 2.300790376298653e-05,
+      "loss": 2.868,
+      "step": 162900
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3050247902024268e-05,
-      "loss": 2.7469,
-      "step": 151000
+      "learning_rate": 2.299133402926215e-05,
+      "loss": 2.8658,
+      "step": 163000
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3032400384078586e-05,
-      "loss": 2.7356,
-      "step": 151100
+      "learning_rate": 2.297476429553777e-05,
+      "loss": 2.8678,
+      "step": 163100
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.3014552866132907e-05,
-      "loss": 2.7421,
-      "step": 151200
+      "learning_rate": 2.2958194561813393e-05,
+      "loss": 2.8703,
+      "step": 163200
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2996705348187228e-05,
-      "loss": 2.7388,
-      "step": 151300
+      "learning_rate": 2.2941624828089014e-05,
+      "loss": 2.8765,
+      "step": 163300
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.297885783024155e-05,
-      "loss": 2.7337,
-      "step": 151400
+      "learning_rate": 2.2925055094364632e-05,
+      "loss": 2.8705,
+      "step": 163400
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.296101031229587e-05,
-      "loss": 2.74,
-      "step": 151500
+      "learning_rate": 2.2908485360640254e-05,
+      "loss": 2.8584,
+      "step": 163500
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.294316279435019e-05,
-      "loss": 2.7336,
-      "step": 151600
+      "learning_rate": 2.289191562691588e-05,
+      "loss": 2.893,
+      "step": 163600
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.292531527640451e-05,
-      "loss": 2.7438,
-      "step": 151700
+      "learning_rate": 2.2875345893191497e-05,
+      "loss": 2.8771,
+      "step": 163700
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.290746775845883e-05,
-      "loss": 2.7376,
-      "step": 151800
+      "learning_rate": 2.285877615946712e-05,
+      "loss": 2.8689,
+      "step": 163800
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2889620240513152e-05,
-      "loss": 2.7321,
-      "step": 151900
+      "learning_rate": 2.284220642574274e-05,
+      "loss": 2.8865,
+      "step": 163900
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2871772722567477e-05,
-      "loss": 2.7423,
-      "step": 152000
+      "learning_rate": 2.2825636692018358e-05,
+      "loss": 2.8772,
+      "step": 164000
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2853925204621794e-05,
-      "loss": 2.7312,
-      "step": 152100
+      "learning_rate": 2.2809066958293983e-05,
+      "loss": 2.8741,
+      "step": 164100
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2836077686676116e-05,
-      "loss": 2.7442,
-      "step": 152200
+      "learning_rate": 2.2792497224569604e-05,
+      "loss": 2.878,
+      "step": 164200
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2818230168730437e-05,
-      "loss": 2.7308,
-      "step": 152300
+      "learning_rate": 2.2775927490845222e-05,
+      "loss": 2.8729,
+      "step": 164300
     },
     {
       "epoch": 0.54,
-      "learning_rate": 2.2800382650784758e-05,
-      "loss": 2.7277,
-      "step": 152400
+      "learning_rate": 2.2759357757120844e-05,
+      "loss": 2.8494,
+      "step": 164400
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 2.278253513283908e-05,
-      "loss": 2.7335,
-      "step": 152500
+      "epoch": 0.55,
+      "learning_rate": 2.2742788023396465e-05,
+      "loss": 2.8626,
+      "step": 164500
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 2.2764687614893397e-05,
-      "loss": 2.7382,
-      "step": 152600
+      "epoch": 0.55,
+      "learning_rate": 2.2726218289672087e-05,
+      "loss": 2.8729,
+      "step": 164600
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2746840096947718e-05,
-      "loss": 2.7324,
-      "step": 152700
+      "learning_rate": 2.2709648555947708e-05,
+      "loss": 2.862,
+      "step": 164700
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.272899257900204e-05,
-      "loss": 2.7361,
-      "step": 152800
+      "learning_rate": 2.269307882222333e-05,
+      "loss": 2.8803,
+      "step": 164800
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.271114506105636e-05,
-      "loss": 2.7301,
-      "step": 152900
+      "learning_rate": 2.2676509088498948e-05,
+      "loss": 2.8778,
+      "step": 164900
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2693297543110682e-05,
-      "loss": 2.7327,
-      "step": 153000
+      "learning_rate": 2.265993935477457e-05,
+      "loss": 2.8776,
+      "step": 165000
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2675450025165003e-05,
-      "loss": 2.7254,
-      "step": 153100
+      "learning_rate": 2.264336962105019e-05,
+      "loss": 2.8631,
+      "step": 165100
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.265760250721932e-05,
-      "loss": 2.7393,
-      "step": 153200
+      "learning_rate": 2.2626799887325812e-05,
+      "loss": 2.876,
+      "step": 165200
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2639754989273642e-05,
-      "loss": 2.7282,
-      "step": 153300
+      "learning_rate": 2.2610230153601434e-05,
+      "loss": 2.8634,
+      "step": 165300
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2621907471327963e-05,
-      "loss": 2.7323,
-      "step": 153400
+      "learning_rate": 2.259366041987705e-05,
+      "loss": 2.8763,
+      "step": 165400
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2604059953382285e-05,
-      "loss": 2.7472,
-      "step": 153500
+      "learning_rate": 2.2577090686152673e-05,
+      "loss": 2.8771,
+      "step": 165500
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2586212435436606e-05,
-      "loss": 2.7348,
-      "step": 153600
+      "learning_rate": 2.2560520952428295e-05,
+      "loss": 2.8611,
+      "step": 165600
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2568364917490924e-05,
-      "loss": 2.7405,
-      "step": 153700
+      "learning_rate": 2.2543951218703916e-05,
+      "loss": 2.8713,
+      "step": 165700
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2550517399545245e-05,
-      "loss": 2.7285,
-      "step": 153800
+      "learning_rate": 2.2527381484979538e-05,
+      "loss": 2.8667,
+      "step": 165800
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2532669881599566e-05,
-      "loss": 2.7314,
-      "step": 153900
+      "learning_rate": 2.251081175125516e-05,
+      "loss": 2.8727,
+      "step": 165900
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2514822363653887e-05,
-      "loss": 2.737,
-      "step": 154000
+      "learning_rate": 2.2494242017530777e-05,
+      "loss": 2.8778,
+      "step": 166000
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.249697484570821e-05,
-      "loss": 2.7351,
-      "step": 154100
+      "learning_rate": 2.24776722838064e-05,
+      "loss": 2.8748,
+      "step": 166100
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.247912732776253e-05,
-      "loss": 2.7373,
-      "step": 154200
+      "learning_rate": 2.2461102550082023e-05,
+      "loss": 2.8804,
+      "step": 166200
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2461279809816848e-05,
-      "loss": 2.7375,
-      "step": 154300
+      "learning_rate": 2.244453281635764e-05,
+      "loss": 2.8674,
+      "step": 166300
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.244343229187117e-05,
-      "loss": 2.7319,
-      "step": 154400
+      "learning_rate": 2.2427963082633263e-05,
+      "loss": 2.8805,
+      "step": 166400
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2425584773925493e-05,
-      "loss": 2.7364,
-      "step": 154500
+      "learning_rate": 2.2411393348908884e-05,
+      "loss": 2.8796,
+      "step": 166500
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2407737255979815e-05,
-      "loss": 2.74,
-      "step": 154600
+      "learning_rate": 2.2394823615184506e-05,
+      "loss": 2.891,
+      "step": 166600
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2389889738034132e-05,
-      "loss": 2.7392,
-      "step": 154700
+      "learning_rate": 2.2378253881460127e-05,
+      "loss": 2.8714,
+      "step": 166700
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2372042220088454e-05,
-      "loss": 2.7344,
-      "step": 154800
+      "learning_rate": 2.236168414773575e-05,
+      "loss": 2.8822,
+      "step": 166800
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2354194702142775e-05,
-      "loss": 2.7385,
-      "step": 154900
+      "learning_rate": 2.2345114414011367e-05,
+      "loss": 2.8691,
+      "step": 166900
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2336347184197096e-05,
-      "loss": 2.7381,
-      "step": 155000
+      "learning_rate": 2.232854468028699e-05,
+      "loss": 2.8791,
+      "step": 167000
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2318499666251417e-05,
-      "loss": 2.7338,
-      "step": 155100
+      "learning_rate": 2.231197494656261e-05,
+      "loss": 2.8758,
+      "step": 167100
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2300652148305735e-05,
-      "loss": 2.735,
-      "step": 155200
+      "learning_rate": 2.229540521283823e-05,
+      "loss": 2.8776,
+      "step": 167200
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2282804630360056e-05,
-      "loss": 2.7421,
-      "step": 155300
+      "learning_rate": 2.2278835479113853e-05,
+      "loss": 2.8943,
+      "step": 167300
     },
     {
       "epoch": 0.55,
-      "learning_rate": 2.2264957112414378e-05,
-      "loss": 2.7342,
-      "step": 155400
+      "learning_rate": 2.2262265745389474e-05,
+      "loss": 2.8668,
+      "step": 167400
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.22471095944687e-05,
-      "loss": 2.7246,
-      "step": 155500
+      "learning_rate": 2.2245696011665092e-05,
+      "loss": 2.8798,
+      "step": 167500
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.222926207652302e-05,
-      "loss": 2.7373,
-      "step": 155600
+      "learning_rate": 2.2229126277940714e-05,
+      "loss": 2.8631,
+      "step": 167600
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.221141455857734e-05,
-      "loss": 2.7279,
-      "step": 155700
+      "learning_rate": 2.2212556544216335e-05,
+      "loss": 2.8719,
+      "step": 167700
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.219356704063166e-05,
-      "loss": 2.7401,
-      "step": 155800
+      "learning_rate": 2.2195986810491957e-05,
+      "loss": 2.8707,
+      "step": 167800
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.217571952268598e-05,
-      "loss": 2.7328,
-      "step": 155900
+      "learning_rate": 2.2179417076767578e-05,
+      "loss": 2.8639,
+      "step": 167900
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.21578720047403e-05,
-      "loss": 2.7382,
-      "step": 156000
+      "learning_rate": 2.2162847343043196e-05,
+      "loss": 2.8705,
+      "step": 168000
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2140024486794623e-05,
-      "loss": 2.7401,
-      "step": 156100
+      "learning_rate": 2.2146277609318818e-05,
+      "loss": 2.8709,
+      "step": 168100
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2122176968848944e-05,
-      "loss": 2.7447,
-      "step": 156200
+      "learning_rate": 2.2129707875594443e-05,
+      "loss": 2.8719,
+      "step": 168200
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2104329450903262e-05,
-      "loss": 2.7364,
-      "step": 156300
+      "learning_rate": 2.211313814187006e-05,
+      "loss": 2.8801,
+      "step": 168300
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2086481932957583e-05,
-      "loss": 2.7355,
-      "step": 156400
+      "learning_rate": 2.2096568408145682e-05,
+      "loss": 2.8804,
+      "step": 168400
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2068634415011904e-05,
-      "loss": 2.7373,
-      "step": 156500
+      "learning_rate": 2.2079998674421304e-05,
+      "loss": 2.8638,
+      "step": 168500
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2050786897066225e-05,
-      "loss": 2.7383,
-      "step": 156600
+      "learning_rate": 2.206342894069692e-05,
+      "loss": 2.8643,
+      "step": 168600
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2032939379120547e-05,
-      "loss": 2.7296,
-      "step": 156700
+      "learning_rate": 2.2046859206972547e-05,
+      "loss": 2.8723,
+      "step": 168700
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.2015091861174868e-05,
-      "loss": 2.742,
-      "step": 156800
+      "learning_rate": 2.2030289473248168e-05,
+      "loss": 2.8526,
+      "step": 168800
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1997244343229186e-05,
-      "loss": 2.7389,
-      "step": 156900
+      "learning_rate": 2.2013719739523786e-05,
+      "loss": 2.8601,
+      "step": 168900
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1979396825283507e-05,
-      "loss": 2.7347,
-      "step": 157000
+      "learning_rate": 2.1997150005799408e-05,
+      "loss": 2.8714,
+      "step": 169000
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.196154930733783e-05,
-      "loss": 2.7319,
-      "step": 157100
+      "learning_rate": 2.198058027207503e-05,
+      "loss": 2.8508,
+      "step": 169100
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1943701789392153e-05,
-      "loss": 2.7376,
-      "step": 157200
+      "learning_rate": 2.196401053835065e-05,
+      "loss": 2.8719,
+      "step": 169200
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.192585427144647e-05,
-      "loss": 2.7379,
-      "step": 157300
+      "learning_rate": 2.1947440804626272e-05,
+      "loss": 2.8757,
+      "step": 169300
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.190800675350079e-05,
-      "loss": 2.7259,
-      "step": 157400
+      "learning_rate": 2.1930871070901893e-05,
+      "loss": 2.8765,
+      "step": 169400
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1890159235555113e-05,
-      "loss": 2.7286,
-      "step": 157500
+      "learning_rate": 2.191430133717751e-05,
+      "loss": 2.8865,
+      "step": 169500
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1872311717609434e-05,
-      "loss": 2.7355,
-      "step": 157600
+      "learning_rate": 2.19433838661223e-05,
+      "loss": 2.8646,
+      "step": 169600
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1854464199663755e-05,
-      "loss": 2.7356,
-      "step": 157700
+      "learning_rate": 2.1926841050005624e-05,
+      "loss": 2.8579,
+      "step": 169700
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1836616681718077e-05,
-      "loss": 2.74,
-      "step": 157800
+      "learning_rate": 2.1910298233888952e-05,
+      "loss": 2.8649,
+      "step": 169800
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1818769163772394e-05,
-      "loss": 2.7324,
-      "step": 157900
+      "learning_rate": 2.189375541777228e-05,
+      "loss": 2.8758,
+      "step": 169900
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1800921645826716e-05,
-      "loss": 2.7417,
-      "step": 158000
+      "learning_rate": 2.1877212601655604e-05,
+      "loss": 2.8744,
+      "step": 170000
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1783074127881037e-05,
-      "loss": 2.7372,
-      "step": 158100
+      "learning_rate": 2.1860669785538935e-05,
+      "loss": 2.8765,
+      "step": 170100
     },
     {
       "epoch": 0.56,
-      "learning_rate": 2.1765226609935358e-05,
-      "loss": 2.7353,
-      "step": 158200
+      "learning_rate": 2.184412696942226e-05,
+      "loss": 2.8615,
+      "step": 170200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1827584153305585e-05,
+      "loss": 2.869,
+      "step": 170300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1811041337188916e-05,
+      "loss": 2.8538,
+      "step": 170400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.179449852107224e-05,
+      "loss": 2.8655,
+      "step": 170500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1777955704955565e-05,
+      "loss": 2.8764,
+      "step": 170600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1761412888838896e-05,
+      "loss": 2.8627,
+      "step": 170700
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.174737909198968e-05,
-      "loss": 2.7333,
-      "step": 158300
+      "learning_rate": 2.174487007272222e-05,
+      "loss": 2.8771,
+      "step": 170800
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1729531574043997e-05,
-      "loss": 2.7282,
-      "step": 158400
+      "learning_rate": 2.1728327256605545e-05,
+      "loss": 2.8683,
+      "step": 170900
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1711684056098318e-05,
-      "loss": 2.7233,
-      "step": 158500
+      "learning_rate": 2.1711784440488873e-05,
+      "loss": 2.8614,
+      "step": 171000
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.169383653815264e-05,
-      "loss": 2.7364,
-      "step": 158600
+      "learning_rate": 2.16952416243722e-05,
+      "loss": 2.8712,
+      "step": 171100
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.167598902020696e-05,
-      "loss": 2.7266,
-      "step": 158700
+      "learning_rate": 2.167869880825553e-05,
+      "loss": 2.8676,
+      "step": 171200
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1658141502261282e-05,
-      "loss": 2.7387,
-      "step": 158800
+      "learning_rate": 2.1662155992138854e-05,
+      "loss": 2.8663,
+      "step": 171300
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1640293984315603e-05,
-      "loss": 2.7276,
-      "step": 158900
+      "learning_rate": 2.164561317602218e-05,
+      "loss": 2.8654,
+      "step": 171400
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.162244646636992e-05,
-      "loss": 2.731,
-      "step": 159000
+      "learning_rate": 2.162907035990551e-05,
+      "loss": 2.8905,
+      "step": 171500
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1604598948424242e-05,
-      "loss": 2.7318,
-      "step": 159100
+      "learning_rate": 2.1612527543788834e-05,
+      "loss": 2.8678,
+      "step": 171600
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1586751430478563e-05,
-      "loss": 2.7236,
-      "step": 159200
+      "learning_rate": 2.1595984727672162e-05,
+      "loss": 2.8592,
+      "step": 171700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.157944191155549e-05,
+      "loss": 2.8708,
+      "step": 171800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1562899095438814e-05,
+      "loss": 2.8686,
+      "step": 171900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1546356279322142e-05,
+      "loss": 2.8865,
+      "step": 172000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.152981346320547e-05,
+      "loss": 2.8724,
+      "step": 172100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1513270647088795e-05,
+      "loss": 2.8729,
+      "step": 172200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1496727830972123e-05,
+      "loss": 2.867,
+      "step": 172300
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1568903912532885e-05,
-      "loss": 2.7312,
-      "step": 159300
+      "learning_rate": 2.148018501485545e-05,
+      "loss": 2.8754,
+      "step": 172400
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1551056394587206e-05,
-      "loss": 2.731,
-      "step": 159400
+      "learning_rate": 2.1463642198738775e-05,
+      "loss": 2.8728,
+      "step": 172500
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1533208876641524e-05,
-      "loss": 2.731,
-      "step": 159500
+      "learning_rate": 2.1447099382622103e-05,
+      "loss": 2.8683,
+      "step": 172600
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1515361358695848e-05,
-      "loss": 2.7295,
-      "step": 159600
+      "learning_rate": 2.143055656650543e-05,
+      "loss": 2.874,
+      "step": 172700
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.149751384075017e-05,
-      "loss": 2.7353,
-      "step": 159700
+      "learning_rate": 2.1414013750388755e-05,
+      "loss": 2.8807,
+      "step": 172800
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.147966632280449e-05,
-      "loss": 2.7237,
-      "step": 159800
+      "learning_rate": 2.1397470934272083e-05,
+      "loss": 2.8749,
+      "step": 172900
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1461818804858812e-05,
-      "loss": 2.7234,
-      "step": 159900
+      "learning_rate": 2.138092811815541e-05,
+      "loss": 2.8732,
+      "step": 173000
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.144397128691313e-05,
-      "loss": 2.7391,
-      "step": 160000
+      "learning_rate": 2.1364385302038736e-05,
+      "loss": 2.8674,
+      "step": 173100
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.142612376896745e-05,
-      "loss": 2.7354,
-      "step": 160100
+      "learning_rate": 2.1347842485922064e-05,
+      "loss": 2.861,
+      "step": 173200
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1408276251021772e-05,
-      "loss": 2.7322,
-      "step": 160200
+      "learning_rate": 2.1331299669805392e-05,
+      "loss": 2.8636,
+      "step": 173300
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1390428733076093e-05,
-      "loss": 2.734,
-      "step": 160300
+      "learning_rate": 2.131475685368872e-05,
+      "loss": 2.8766,
+      "step": 173400
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1372581215130415e-05,
-      "loss": 2.7288,
-      "step": 160400
+      "learning_rate": 2.1298214037572044e-05,
+      "loss": 2.882,
+      "step": 173500
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1354733697184732e-05,
-      "loss": 2.7248,
-      "step": 160500
+      "learning_rate": 2.1281671221455372e-05,
+      "loss": 2.8712,
+      "step": 173600
     },
     {
       "epoch": 0.57,
-      "learning_rate": 2.1336886179239054e-05,
-      "loss": 2.7324,
-      "step": 160600
+      "learning_rate": 2.12651284053387e-05,
+      "loss": 2.8789,
+      "step": 173700
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 2.1319038661293375e-05,
-      "loss": 2.7315,
-      "step": 160700
+      "epoch": 0.58,
+      "learning_rate": 2.1248585589222025e-05,
+      "loss": 2.8739,
+      "step": 173800
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 2.1301191143347696e-05,
-      "loss": 2.7247,
-      "step": 160800
+      "epoch": 0.58,
+      "learning_rate": 2.1232042773105352e-05,
+      "loss": 2.881,
+      "step": 173900
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 2.1283343625402017e-05,
-      "loss": 2.7203,
-      "step": 160900
+      "epoch": 0.58,
+      "learning_rate": 2.121549995698868e-05,
+      "loss": 2.869,
+      "step": 174000
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 2.126549610745634e-05,
-      "loss": 2.7303,
-      "step": 161000
+      "epoch": 0.58,
+      "learning_rate": 2.1198957140872005e-05,
+      "loss": 2.8644,
+      "step": 174100
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1247648589510656e-05,
-      "loss": 2.7359,
-      "step": 161100
+      "learning_rate": 2.1182414324755333e-05,
+      "loss": 2.873,
+      "step": 174200
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1229801071564978e-05,
-      "loss": 2.7361,
-      "step": 161200
+      "learning_rate": 2.116587150863866e-05,
+      "loss": 2.8757,
+      "step": 174300
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.12119535536193e-05,
-      "loss": 2.7247,
-      "step": 161300
+      "learning_rate": 2.1149328692521985e-05,
+      "loss": 2.8631,
+      "step": 174400
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.119410603567362e-05,
-      "loss": 2.7246,
-      "step": 161400
+      "learning_rate": 2.1132785876405313e-05,
+      "loss": 2.8707,
+      "step": 174500
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.117625851772794e-05,
-      "loss": 2.7382,
-      "step": 161500
+      "learning_rate": 2.111624306028864e-05,
+      "loss": 2.8743,
+      "step": 174600
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.115841099978226e-05,
-      "loss": 2.7319,
-      "step": 161600
+      "learning_rate": 2.1099700244171966e-05,
+      "loss": 2.8778,
+      "step": 174700
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.114056348183658e-05,
-      "loss": 2.7361,
-      "step": 161700
+      "learning_rate": 2.1083157428055294e-05,
+      "loss": 2.8802,
+      "step": 174800
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.11227159638909e-05,
-      "loss": 2.7345,
-      "step": 161800
+      "learning_rate": 2.106661461193862e-05,
+      "loss": 2.879,
+      "step": 174900
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1104868445945223e-05,
-      "loss": 2.7344,
-      "step": 161900
+      "learning_rate": 2.1050071795821946e-05,
+      "loss": 2.8592,
+      "step": 175000
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1087020927999544e-05,
-      "loss": 2.7123,
-      "step": 162000
+      "learning_rate": 2.1033528979705274e-05,
+      "loss": 2.8658,
+      "step": 175100
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1069173410053865e-05,
-      "loss": 2.7427,
-      "step": 162100
+      "learning_rate": 2.1016986163588602e-05,
+      "loss": 2.8716,
+      "step": 175200
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1051325892108186e-05,
-      "loss": 2.7383,
-      "step": 162200
+      "learning_rate": 2.100044334747193e-05,
+      "loss": 2.8791,
+      "step": 175300
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.1033478374162507e-05,
-      "loss": 2.7346,
-      "step": 162300
+      "learning_rate": 2.0983900531355254e-05,
+      "loss": 2.8672,
+      "step": 175400
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.101563085621683e-05,
-      "loss": 2.7368,
-      "step": 162400
+      "learning_rate": 2.0967357715238582e-05,
+      "loss": 2.8761,
+      "step": 175500
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.099778333827115e-05,
-      "loss": 2.7379,
-      "step": 162500
+      "learning_rate": 2.095081489912191e-05,
+      "loss": 2.8616,
+      "step": 175600
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0979935820325468e-05,
-      "loss": 2.7412,
-      "step": 162600
+      "learning_rate": 2.0934272083005235e-05,
+      "loss": 2.8713,
+      "step": 175700
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.096208830237979e-05,
-      "loss": 2.7277,
-      "step": 162700
+      "learning_rate": 2.0917729266888563e-05,
+      "loss": 2.8649,
+      "step": 175800
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.094424078443411e-05,
-      "loss": 2.7263,
-      "step": 162800
+      "learning_rate": 2.090118645077189e-05,
+      "loss": 2.8763,
+      "step": 175900
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.092639326648843e-05,
-      "loss": 2.743,
-      "step": 162900
+      "learning_rate": 2.0884643634655215e-05,
+      "loss": 2.8719,
+      "step": 176000
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0908545748542753e-05,
-      "loss": 2.7327,
-      "step": 163000
+      "learning_rate": 2.086810081853854e-05,
+      "loss": 2.882,
+      "step": 176100
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0890698230597074e-05,
-      "loss": 2.7256,
-      "step": 163100
+      "learning_rate": 2.085155800242187e-05,
+      "loss": 2.8732,
+      "step": 176200
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.087285071265139e-05,
-      "loss": 2.7262,
-      "step": 163200
+      "learning_rate": 2.0835015186305196e-05,
+      "loss": 2.8809,
+      "step": 176300
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0855003194705713e-05,
-      "loss": 2.7296,
-      "step": 163300
+      "learning_rate": 2.0818472370188523e-05,
+      "loss": 2.8732,
+      "step": 176400
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0837155676760034e-05,
-      "loss": 2.7268,
-      "step": 163400
+      "learning_rate": 2.080192955407185e-05,
+      "loss": 2.8756,
+      "step": 176500
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0819308158814355e-05,
-      "loss": 2.7389,
-      "step": 163500
+      "learning_rate": 2.0785386737955176e-05,
+      "loss": 2.8601,
+      "step": 176600
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0801460640868677e-05,
-      "loss": 2.7382,
-      "step": 163600
+      "learning_rate": 2.0768843921838504e-05,
+      "loss": 2.8599,
+      "step": 176700
     },
     {
       "epoch": 0.58,
-      "learning_rate": 2.0783613122922994e-05,
-      "loss": 2.7265,
-      "step": 163700
+      "learning_rate": 2.0752301105721832e-05,
+      "loss": 2.8808,
+      "step": 176800
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 2.0765765604977316e-05,
-      "loss": 2.7362,
-      "step": 163800
+      "epoch": 0.59,
+      "learning_rate": 2.0735758289605156e-05,
+      "loss": 2.8623,
+      "step": 176900
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0747918087031637e-05,
-      "loss": 2.7381,
-      "step": 163900
+      "learning_rate": 2.0719215473488484e-05,
+      "loss": 2.8659,
+      "step": 177000
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0730070569085958e-05,
-      "loss": 2.7294,
-      "step": 164000
+      "learning_rate": 2.0702672657371812e-05,
+      "loss": 2.8613,
+      "step": 177100
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.071222305114028e-05,
-      "loss": 2.734,
-      "step": 164100
+      "learning_rate": 2.0686129841255137e-05,
+      "loss": 2.8636,
+      "step": 177200
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.06943755331946e-05,
-      "loss": 2.7334,
-      "step": 164200
+      "learning_rate": 2.0669587025138465e-05,
+      "loss": 2.8746,
+      "step": 177300
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0676528015248918e-05,
-      "loss": 2.7351,
-      "step": 164300
+      "learning_rate": 2.065304420902179e-05,
+      "loss": 2.8714,
+      "step": 177400
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.065868049730324e-05,
-      "loss": 2.7395,
-      "step": 164400
+      "learning_rate": 2.063650139290512e-05,
+      "loss": 2.8696,
+      "step": 177500
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.064083297935756e-05,
-      "loss": 2.7273,
-      "step": 164500
+      "learning_rate": 2.0619958576788445e-05,
+      "loss": 2.8745,
+      "step": 177600
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0622985461411882e-05,
-      "loss": 2.7393,
-      "step": 164600
+      "learning_rate": 2.060341576067177e-05,
+      "loss": 2.8743,
+      "step": 177700
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0605137943466203e-05,
-      "loss": 2.7142,
-      "step": 164700
+      "learning_rate": 2.05868729445551e-05,
+      "loss": 2.8609,
+      "step": 177800
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0587290425520524e-05,
-      "loss": 2.7215,
-      "step": 164800
+      "learning_rate": 2.0570330128438425e-05,
+      "loss": 2.8692,
+      "step": 177900
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0569442907574846e-05,
-      "loss": 2.7487,
-      "step": 164900
+      "learning_rate": 2.055378731232175e-05,
+      "loss": 2.8547,
+      "step": 178000
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0551595389629167e-05,
-      "loss": 2.7345,
-      "step": 165000
+      "learning_rate": 2.053724449620508e-05,
+      "loss": 2.8733,
+      "step": 178100
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0533747871683488e-05,
-      "loss": 2.7379,
-      "step": 165100
+      "learning_rate": 2.0520701680088406e-05,
+      "loss": 2.8692,
+      "step": 178200
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.051590035373781e-05,
-      "loss": 2.7307,
-      "step": 165200
+      "learning_rate": 2.050415886397173e-05,
+      "loss": 2.8784,
+      "step": 178300
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0498052835792127e-05,
-      "loss": 2.7303,
-      "step": 165300
+      "learning_rate": 2.048761604785506e-05,
+      "loss": 2.8828,
+      "step": 178400
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0480205317846448e-05,
-      "loss": 2.7239,
-      "step": 165400
+      "learning_rate": 2.0471073231738386e-05,
+      "loss": 2.857,
+      "step": 178500
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.046235779990077e-05,
-      "loss": 2.7309,
-      "step": 165500
+      "learning_rate": 2.0454530415621714e-05,
+      "loss": 2.8672,
+      "step": 178600
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.044451028195509e-05,
-      "loss": 2.7322,
-      "step": 165600
+      "learning_rate": 2.043798759950504e-05,
+      "loss": 2.8747,
+      "step": 178700
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0426662764009412e-05,
-      "loss": 2.733,
-      "step": 165700
+      "learning_rate": 2.0421444783388366e-05,
+      "loss": 2.865,
+      "step": 178800
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.040881524606373e-05,
-      "loss": 2.7296,
-      "step": 165800
+      "learning_rate": 2.0404901967271694e-05,
+      "loss": 2.8639,
+      "step": 178900
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.039096772811805e-05,
-      "loss": 2.7305,
-      "step": 165900
+      "learning_rate": 2.038835915115502e-05,
+      "loss": 2.8654,
+      "step": 179000
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0373120210172372e-05,
-      "loss": 2.7323,
-      "step": 166000
+      "learning_rate": 2.0371816335038347e-05,
+      "loss": 2.8556,
+      "step": 179100
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0355272692226693e-05,
-      "loss": 2.7186,
-      "step": 166100
+      "learning_rate": 2.0355273518921675e-05,
+      "loss": 2.8737,
+      "step": 179200
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0337425174281015e-05,
-      "loss": 2.7257,
-      "step": 166200
+      "learning_rate": 2.0338730702805e-05,
+      "loss": 2.8759,
+      "step": 179300
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0319577656335332e-05,
-      "loss": 2.724,
-      "step": 166300
+      "learning_rate": 2.0322187886688327e-05,
+      "loss": 2.8637,
+      "step": 179400
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0301730138389654e-05,
-      "loss": 2.744,
-      "step": 166400
+      "learning_rate": 2.0305645070571655e-05,
+      "loss": 2.8625,
+      "step": 179500
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0283882620443975e-05,
-      "loss": 2.7325,
-      "step": 166500
+      "learning_rate": 2.028910225445498e-05,
+      "loss": 2.8789,
+      "step": 179600
     },
     {
       "epoch": 0.59,
-      "learning_rate": 2.0266035102498296e-05,
-      "loss": 2.7355,
-      "step": 166600
+      "learning_rate": 2.027255943833831e-05,
+      "loss": 2.8696,
+      "step": 179700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0256016622221636e-05,
+      "loss": 2.8672,
+      "step": 179800
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0248187584552617e-05,
-      "loss": 2.7348,
-      "step": 166700
+      "learning_rate": 2.023947380610496e-05,
+      "loss": 2.8693,
+      "step": 179900
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.023034006660694e-05,
-      "loss": 2.7232,
-      "step": 166800
+      "learning_rate": 2.0222930989988288e-05,
+      "loss": 2.8672,
+      "step": 180000
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0212492548661256e-05,
-      "loss": 2.7342,
-      "step": 166900
+      "learning_rate": 2.0206388173871616e-05,
+      "loss": 2.8805,
+      "step": 180100
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0194645030715577e-05,
-      "loss": 2.736,
-      "step": 167000
+      "learning_rate": 2.018984535775494e-05,
+      "loss": 2.866,
+      "step": 180200
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.01767975127699e-05,
-      "loss": 2.7335,
-      "step": 167100
+      "learning_rate": 2.017330254163827e-05,
+      "loss": 2.8533,
+      "step": 180300
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.015894999482422e-05,
-      "loss": 2.7272,
-      "step": 167200
+      "learning_rate": 2.0156759725521596e-05,
+      "loss": 2.8616,
+      "step": 180400
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.014110247687854e-05,
-      "loss": 2.7345,
-      "step": 167300
+      "learning_rate": 2.014021690940492e-05,
+      "loss": 2.8675,
+      "step": 180500
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0123254958932862e-05,
-      "loss": 2.7304,
-      "step": 167400
+      "learning_rate": 2.012367409328825e-05,
+      "loss": 2.8803,
+      "step": 180600
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0105407440987184e-05,
-      "loss": 2.7237,
-      "step": 167500
+      "learning_rate": 2.0107131277171577e-05,
+      "loss": 2.8723,
+      "step": 180700
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0087559923041505e-05,
-      "loss": 2.7293,
-      "step": 167600
+      "learning_rate": 2.0090588461054905e-05,
+      "loss": 2.8735,
+      "step": 180800
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0069712405095826e-05,
-      "loss": 2.7264,
-      "step": 167700
+      "learning_rate": 2.007404564493823e-05,
+      "loss": 2.8867,
+      "step": 180900
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0051864887150147e-05,
-      "loss": 2.7075,
-      "step": 167800
+      "learning_rate": 2.0057502828821557e-05,
+      "loss": 2.8691,
+      "step": 181000
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0034017369204465e-05,
-      "loss": 2.736,
-      "step": 167900
+      "learning_rate": 2.0040960012704885e-05,
+      "loss": 2.8553,
+      "step": 181100
     },
     {
       "epoch": 0.6,
-      "learning_rate": 2.0016169851258786e-05,
-      "loss": 2.7302,
-      "step": 168000
+      "learning_rate": 2.002441719658821e-05,
+      "loss": 2.8684,
+      "step": 181200
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.9998322333313107e-05,
-      "loss": 2.7386,
-      "step": 168100
+      "learning_rate": 2.0007874380471537e-05,
+      "loss": 2.8764,
+      "step": 181300
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.998047481536743e-05,
-      "loss": 2.7324,
-      "step": 168200
+      "learning_rate": 1.9991331564354865e-05,
+      "loss": 2.862,
+      "step": 181400
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.996262729742175e-05,
-      "loss": 2.7342,
-      "step": 168300
+      "learning_rate": 1.997478874823819e-05,
+      "loss": 2.8726,
+      "step": 181500
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.9944779779476068e-05,
-      "loss": 2.7263,
-      "step": 168400
+      "learning_rate": 1.9958245932121518e-05,
+      "loss": 2.8695,
+      "step": 181600
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.992693226153039e-05,
-      "loss": 2.728,
-      "step": 168500
+      "learning_rate": 1.9941703116004846e-05,
+      "loss": 2.8722,
+      "step": 181700
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.990908474358471e-05,
-      "loss": 2.7226,
-      "step": 168600
+      "learning_rate": 1.992516029988817e-05,
+      "loss": 2.8641,
+      "step": 181800
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.989123722563903e-05,
-      "loss": 2.7231,
-      "step": 168700
+      "learning_rate": 1.9908617483771498e-05,
+      "loss": 2.8634,
+      "step": 181900
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.9873389707693353e-05,
-      "loss": 2.7286,
-      "step": 168800
+      "learning_rate": 1.9892074667654826e-05,
+      "loss": 2.8701,
+      "step": 182000
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.9855542189747674e-05,
-      "loss": 2.7298,
-      "step": 168900
+      "learning_rate": 1.987553185153815e-05,
+      "loss": 2.8728,
+      "step": 182100
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.983769467180199e-05,
-      "loss": 2.727,
-      "step": 169000
+      "learning_rate": 1.985898903542148e-05,
+      "loss": 2.8634,
+      "step": 182200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9842446219304807e-05,
+      "loss": 2.8709,
+      "step": 182300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.982590340318813e-05,
+      "loss": 2.8731,
+      "step": 182400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980936058707146e-05,
+      "loss": 2.8624,
+      "step": 182500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9792817770954787e-05,
+      "loss": 2.8702,
+      "step": 182600
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.9819847153856313e-05,
-      "loss": 2.728,
-      "step": 169100
+      "learning_rate": 1.9776274954838115e-05,
+      "loss": 2.8558,
+      "step": 182700
     },
     {
       "epoch": 0.6,
-      "learning_rate": 1.9801999635910634e-05,
-      "loss": 2.7248,
-      "step": 169200
+      "learning_rate": 1.975973213872144e-05,
+      "loss": 2.8703,
+      "step": 182800
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.9784152117964955e-05,
-      "loss": 2.7209,
-      "step": 169300
+      "epoch": 0.61,
+      "learning_rate": 1.9743189322604767e-05,
+      "loss": 2.8744,
+      "step": 182900
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.9766304600019276e-05,
-      "loss": 2.7347,
-      "step": 169400
+      "epoch": 0.61,
+      "learning_rate": 1.9726646506488095e-05,
+      "loss": 2.8597,
+      "step": 183000
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9748457082073594e-05,
-      "loss": 2.723,
-      "step": 169500
+      "learning_rate": 1.971010369037142e-05,
+      "loss": 2.8806,
+      "step": 183100
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9730609564127916e-05,
-      "loss": 2.7269,
-      "step": 169600
+      "learning_rate": 1.9693560874254748e-05,
+      "loss": 2.8724,
+      "step": 183200
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9712762046182237e-05,
-      "loss": 2.7281,
-      "step": 169700
+      "learning_rate": 1.9677018058138076e-05,
+      "loss": 2.8759,
+      "step": 183300
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.969491452823656e-05,
-      "loss": 2.7226,
-      "step": 169800
+      "learning_rate": 1.96604752420214e-05,
+      "loss": 2.8819,
+      "step": 183400
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9677067010290883e-05,
-      "loss": 2.7264,
-      "step": 169900
+      "learning_rate": 1.9643932425904728e-05,
+      "loss": 2.8727,
+      "step": 183500
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.96592194923452e-05,
-      "loss": 2.7239,
-      "step": 170000
+      "learning_rate": 1.9627389609788056e-05,
+      "loss": 2.8578,
+      "step": 183600
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.964137197439952e-05,
-      "loss": 2.7346,
-      "step": 170100
+      "learning_rate": 1.961084679367138e-05,
+      "loss": 2.8695,
+      "step": 183700
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9623524456453843e-05,
-      "loss": 2.7238,
-      "step": 170200
+      "learning_rate": 1.959430397755471e-05,
+      "loss": 2.861,
+      "step": 183800
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9605676938508164e-05,
-      "loss": 2.7189,
-      "step": 170300
+      "learning_rate": 1.9577761161438036e-05,
+      "loss": 2.8655,
+      "step": 183900
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9587829420562485e-05,
-      "loss": 2.71,
-      "step": 170400
+      "learning_rate": 1.956121834532136e-05,
+      "loss": 2.8686,
+      "step": 184000
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9569981902616803e-05,
-      "loss": 2.7351,
-      "step": 170500
+      "learning_rate": 1.954467552920469e-05,
+      "loss": 2.8721,
+      "step": 184100
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9552134384671124e-05,
-      "loss": 2.7317,
-      "step": 170600
+      "learning_rate": 1.9528132713088017e-05,
+      "loss": 2.8587,
+      "step": 184200
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9534286866725445e-05,
-      "loss": 2.7294,
-      "step": 170700
+      "learning_rate": 1.951158989697134e-05,
+      "loss": 2.8789,
+      "step": 184300
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9516439348779767e-05,
-      "loss": 2.7275,
-      "step": 170800
+      "learning_rate": 1.949504708085467e-05,
+      "loss": 2.8628,
+      "step": 184400
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9498591830834088e-05,
-      "loss": 2.7257,
-      "step": 170900
+      "learning_rate": 1.9478504264737997e-05,
+      "loss": 2.8795,
+      "step": 184500
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.948074431288841e-05,
-      "loss": 2.7292,
-      "step": 171000
+      "learning_rate": 1.946196144862132e-05,
+      "loss": 2.8762,
+      "step": 184600
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9462896794942727e-05,
-      "loss": 2.7256,
-      "step": 171100
+      "learning_rate": 1.944541863250465e-05,
+      "loss": 2.8683,
+      "step": 184700
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9445049276997048e-05,
-      "loss": 2.7299,
-      "step": 171200
+      "learning_rate": 1.9428875816387977e-05,
+      "loss": 2.8684,
+      "step": 184800
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.942720175905137e-05,
-      "loss": 2.7187,
-      "step": 171300
+      "learning_rate": 1.9412333000271305e-05,
+      "loss": 2.8762,
+      "step": 184900
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.940935424110569e-05,
-      "loss": 2.7382,
-      "step": 171400
+      "learning_rate": 1.939579018415463e-05,
+      "loss": 2.8703,
+      "step": 185000
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9391506723160012e-05,
-      "loss": 2.7323,
-      "step": 171500
+      "learning_rate": 1.9379247368037954e-05,
+      "loss": 2.8686,
+      "step": 185100
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.937365920521433e-05,
-      "loss": 2.7233,
-      "step": 171600
+      "learning_rate": 1.9362704551921286e-05,
+      "loss": 2.8725,
+      "step": 185200
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.935581168726865e-05,
-      "loss": 2.7148,
-      "step": 171700
+      "learning_rate": 1.934616173580461e-05,
+      "loss": 2.87,
+      "step": 185300
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9337964169322972e-05,
-      "loss": 2.727,
-      "step": 171800
+      "learning_rate": 1.9329618919687935e-05,
+      "loss": 2.8682,
+      "step": 185400
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9320116651377293e-05,
-      "loss": 2.731,
-      "step": 171900
+      "learning_rate": 1.9313076103571266e-05,
+      "loss": 2.8549,
+      "step": 185500
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9302269133431615e-05,
-      "loss": 2.7285,
-      "step": 172000
+      "learning_rate": 1.929653328745459e-05,
+      "loss": 2.8652,
+      "step": 185600
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9284421615485936e-05,
-      "loss": 2.7351,
-      "step": 172100
+      "learning_rate": 1.9279990471337915e-05,
+      "loss": 2.8668,
+      "step": 185700
     },
     {
       "epoch": 0.61,
-      "learning_rate": 1.9266574097540254e-05,
-      "loss": 2.7242,
-      "step": 172200
+      "learning_rate": 1.9263447655221247e-05,
+      "loss": 2.8621,
+      "step": 185800
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9248726579594575e-05,
-      "loss": 2.7252,
-      "step": 172300
+      "learning_rate": 1.924690483910457e-05,
+      "loss": 2.8517,
+      "step": 185900
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.92308790616489e-05,
-      "loss": 2.7327,
-      "step": 172400
+      "learning_rate": 1.92303620229879e-05,
+      "loss": 2.8601,
+      "step": 186000
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.921303154370322e-05,
-      "loss": 2.7288,
-      "step": 172500
+      "learning_rate": 1.9213819206871227e-05,
+      "loss": 2.8607,
+      "step": 186100
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.919518402575754e-05,
-      "loss": 2.7168,
-      "step": 172600
+      "learning_rate": 1.919727639075455e-05,
+      "loss": 2.8722,
+      "step": 186200
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.917733650781186e-05,
-      "loss": 2.7302,
-      "step": 172700
+      "learning_rate": 1.918073357463788e-05,
+      "loss": 2.8673,
+      "step": 186300
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.915948898986618e-05,
-      "loss": 2.7306,
-      "step": 172800
+      "learning_rate": 1.9164190758521204e-05,
+      "loss": 2.8604,
+      "step": 186400
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9141641471920502e-05,
-      "loss": 2.7166,
-      "step": 172900
+      "learning_rate": 1.9147647942404532e-05,
+      "loss": 2.8584,
+      "step": 186500
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9123793953974823e-05,
-      "loss": 2.7168,
-      "step": 173000
+      "learning_rate": 1.913110512628786e-05,
+      "loss": 2.8667,
+      "step": 186600
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9105946436029144e-05,
-      "loss": 2.736,
-      "step": 173100
+      "learning_rate": 1.9114562310171184e-05,
+      "loss": 2.8684,
+      "step": 186700
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9088098918083462e-05,
-      "loss": 2.724,
-      "step": 173200
+      "learning_rate": 1.9098019494054512e-05,
+      "loss": 2.8575,
+      "step": 186800
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9070251400137784e-05,
-      "loss": 2.7258,
-      "step": 173300
+      "learning_rate": 1.908147667793784e-05,
+      "loss": 2.8651,
+      "step": 186900
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9052403882192105e-05,
-      "loss": 2.723,
-      "step": 173400
+      "learning_rate": 1.9064933861821165e-05,
+      "loss": 2.8593,
+      "step": 187000
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9034556364246426e-05,
-      "loss": 2.7322,
-      "step": 173500
+      "learning_rate": 1.9048391045704496e-05,
+      "loss": 2.8595,
+      "step": 187100
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.9016708846300747e-05,
-      "loss": 2.7307,
-      "step": 173600
+      "learning_rate": 1.903184822958782e-05,
+      "loss": 2.8615,
+      "step": 187200
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8998861328355065e-05,
-      "loss": 2.7214,
-      "step": 173700
+      "learning_rate": 1.9015305413471145e-05,
+      "loss": 2.8587,
+      "step": 187300
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8981013810409386e-05,
-      "loss": 2.7123,
-      "step": 173800
+      "learning_rate": 1.8998762597354476e-05,
+      "loss": 2.8845,
+      "step": 187400
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8963166292463707e-05,
-      "loss": 2.731,
-      "step": 173900
+      "learning_rate": 1.89822197812378e-05,
+      "loss": 2.8724,
+      "step": 187500
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.894531877451803e-05,
-      "loss": 2.7262,
-      "step": 174000
+      "learning_rate": 1.8965676965121125e-05,
+      "loss": 2.8748,
+      "step": 187600
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.892747125657235e-05,
-      "loss": 2.7263,
-      "step": 174100
+      "learning_rate": 1.8949134149004453e-05,
+      "loss": 2.8527,
+      "step": 187700
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.890962373862667e-05,
-      "loss": 2.7317,
-      "step": 174200
+      "learning_rate": 1.893259133288778e-05,
+      "loss": 2.862,
+      "step": 187800
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.889177622068099e-05,
-      "loss": 2.7196,
-      "step": 174300
+      "learning_rate": 1.8916048516771106e-05,
+      "loss": 2.852,
+      "step": 187900
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.887392870273531e-05,
-      "loss": 2.7145,
-      "step": 174400
+      "learning_rate": 1.8899505700654434e-05,
+      "loss": 2.8668,
+      "step": 188000
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.885608118478963e-05,
-      "loss": 2.731,
-      "step": 174500
+      "learning_rate": 1.888296288453776e-05,
+      "loss": 2.8622,
+      "step": 188100
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8838233666843953e-05,
-      "loss": 2.7168,
-      "step": 174600
+      "learning_rate": 1.886642006842109e-05,
+      "loss": 2.8866,
+      "step": 188200
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8820386148898274e-05,
-      "loss": 2.7146,
-      "step": 174700
+      "learning_rate": 1.8849877252304414e-05,
+      "loss": 2.864,
+      "step": 188300
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.880253863095259e-05,
-      "loss": 2.7292,
-      "step": 174800
+      "learning_rate": 1.8833334436187742e-05,
+      "loss": 2.861,
+      "step": 188400
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8784691113006913e-05,
-      "loss": 2.7264,
-      "step": 174900
+      "learning_rate": 1.881679162007107e-05,
+      "loss": 2.8577,
+      "step": 188500
     },
     {
       "epoch": 0.62,
-      "learning_rate": 1.8766843595061237e-05,
-      "loss": 2.7189,
-      "step": 175000
+      "learning_rate": 1.8800248803954394e-05,
+      "loss": 2.8753,
+      "step": 188600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8783705987837722e-05,
+      "loss": 2.8752,
+      "step": 188700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.876716317172105e-05,
+      "loss": 2.8684,
+      "step": 188800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8750620355604375e-05,
+      "loss": 2.8653,
+      "step": 188900
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.874899607711556e-05,
-      "loss": 2.7263,
-      "step": 175100
+      "learning_rate": 1.8734077539487703e-05,
+      "loss": 2.8674,
+      "step": 189000
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.873114855916988e-05,
-      "loss": 2.7316,
-      "step": 175200
+      "learning_rate": 1.871753472337103e-05,
+      "loss": 2.8686,
+      "step": 189100
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8713301041224198e-05,
-      "loss": 2.7149,
-      "step": 175300
+      "learning_rate": 1.8700991907254355e-05,
+      "loss": 2.8555,
+      "step": 189200
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.869545352327852e-05,
-      "loss": 2.7134,
-      "step": 175400
+      "learning_rate": 1.8684449091137683e-05,
+      "loss": 2.8674,
+      "step": 189300
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.867760600533284e-05,
-      "loss": 2.7276,
-      "step": 175500
+      "learning_rate": 1.866790627502101e-05,
+      "loss": 2.8673,
+      "step": 189400
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.865975848738716e-05,
-      "loss": 2.7289,
-      "step": 175600
+      "learning_rate": 1.8651363458904336e-05,
+      "loss": 2.8582,
+      "step": 189500
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8641910969441483e-05,
-      "loss": 2.7301,
-      "step": 175700
+      "learning_rate": 1.8634820642787664e-05,
+      "loss": 2.874,
+      "step": 189600
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.86240634514958e-05,
-      "loss": 2.7229,
-      "step": 175800
+      "learning_rate": 1.861827782667099e-05,
+      "loss": 2.8608,
+      "step": 189700
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.860621593355012e-05,
-      "loss": 2.7207,
-      "step": 175900
+      "learning_rate": 1.8601735010554316e-05,
+      "loss": 2.8664,
+      "step": 189800
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8588368415604443e-05,
-      "loss": 2.7187,
-      "step": 176000
+      "learning_rate": 1.8585192194437644e-05,
+      "loss": 2.8692,
+      "step": 189900
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8570520897658764e-05,
-      "loss": 2.7242,
-      "step": 176100
+      "learning_rate": 1.8568649378320972e-05,
+      "loss": 2.8667,
+      "step": 190000
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8552673379713085e-05,
-      "loss": 2.7331,
-      "step": 176200
+      "learning_rate": 1.85521065622043e-05,
+      "loss": 2.8787,
+      "step": 190100
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8534825861767406e-05,
-      "loss": 2.728,
-      "step": 176300
+      "learning_rate": 1.8535563746087624e-05,
+      "loss": 2.8593,
+      "step": 190200
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8516978343821724e-05,
-      "loss": 2.7247,
-      "step": 176400
+      "learning_rate": 1.8519020929970952e-05,
+      "loss": 2.8659,
+      "step": 190300
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8499130825876045e-05,
-      "loss": 2.7274,
-      "step": 176500
+      "learning_rate": 1.850247811385428e-05,
+      "loss": 2.8526,
+      "step": 190400
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8481283307930367e-05,
-      "loss": 2.7342,
-      "step": 176600
+      "learning_rate": 1.8485935297737605e-05,
+      "loss": 2.8682,
+      "step": 190500
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8463435789984688e-05,
-      "loss": 2.7282,
-      "step": 176700
+      "learning_rate": 1.8469392481620933e-05,
+      "loss": 2.855,
+      "step": 190600
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.844558827203901e-05,
-      "loss": 2.7143,
-      "step": 176800
+      "learning_rate": 1.845284966550426e-05,
+      "loss": 2.8619,
+      "step": 190700
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8427740754093327e-05,
-      "loss": 2.7194,
-      "step": 176900
+      "learning_rate": 1.8436306849387585e-05,
+      "loss": 2.8659,
+      "step": 190800
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8409893236147648e-05,
-      "loss": 2.7324,
-      "step": 177000
+      "learning_rate": 1.8419764033270913e-05,
+      "loss": 2.8698,
+      "step": 190900
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.839204571820197e-05,
-      "loss": 2.7307,
-      "step": 177100
+      "learning_rate": 1.840322121715424e-05,
+      "loss": 2.8539,
+      "step": 191000
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.837419820025629e-05,
-      "loss": 2.7296,
-      "step": 177200
+      "learning_rate": 1.8386678401037565e-05,
+      "loss": 2.8629,
+      "step": 191100
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8356350682310612e-05,
-      "loss": 2.7236,
-      "step": 177300
+      "learning_rate": 1.8370135584920893e-05,
+      "loss": 2.8574,
+      "step": 191200
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.833850316436493e-05,
-      "loss": 2.7156,
-      "step": 177400
+      "learning_rate": 1.835359276880422e-05,
+      "loss": 2.8591,
+      "step": 191300
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8320655646419254e-05,
-      "loss": 2.7224,
-      "step": 177500
+      "learning_rate": 1.8337049952687546e-05,
+      "loss": 2.8584,
+      "step": 191400
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8302808128473575e-05,
-      "loss": 2.7156,
-      "step": 177600
+      "learning_rate": 1.8320507136570874e-05,
+      "loss": 2.8731,
+      "step": 191500
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8284960610527897e-05,
-      "loss": 2.724,
-      "step": 177700
+      "learning_rate": 1.83039643204542e-05,
+      "loss": 2.8758,
+      "step": 191600
     },
     {
       "epoch": 0.63,
-      "learning_rate": 1.8267113092582218e-05,
-      "loss": 2.7244,
-      "step": 177800
+      "learning_rate": 1.8287421504337526e-05,
+      "loss": 2.86,
+      "step": 191700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8270878688220854e-05,
+      "loss": 2.8567,
+      "step": 191800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8254335872104182e-05,
+      "loss": 2.8548,
+      "step": 191900
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8249265574636536e-05,
-      "loss": 2.7247,
-      "step": 177900
+      "learning_rate": 1.8237793055987507e-05,
+      "loss": 2.8604,
+      "step": 192000
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8231418056690857e-05,
-      "loss": 2.7214,
-      "step": 178000
+      "learning_rate": 1.8221250239870835e-05,
+      "loss": 2.8635,
+      "step": 192100
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8213570538745178e-05,
-      "loss": 2.7257,
-      "step": 178100
+      "learning_rate": 1.8204707423754162e-05,
+      "loss": 2.8641,
+      "step": 192200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.818816460763749e-05,
+      "loss": 2.8747,
+      "step": 192300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8171621791520815e-05,
+      "loss": 2.8531,
+      "step": 192400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8155078975404143e-05,
+      "loss": 2.859,
+      "step": 192500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.813853615928747e-05,
+      "loss": 2.8526,
+      "step": 192600
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.81957230207995e-05,
-      "loss": 2.7253,
-      "step": 178200
+      "learning_rate": 1.8121993343170795e-05,
+      "loss": 2.8534,
+      "step": 192700
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.817787550285382e-05,
-      "loss": 2.7244,
-      "step": 178300
+      "learning_rate": 1.810545052705412e-05,
+      "loss": 2.8638,
+      "step": 192800
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.816002798490814e-05,
-      "loss": 2.7392,
-      "step": 178400
+      "learning_rate": 1.808890771093745e-05,
+      "loss": 2.8628,
+      "step": 192900
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.814218046696246e-05,
-      "loss": 2.7217,
-      "step": 178500
+      "learning_rate": 1.8072364894820776e-05,
+      "loss": 2.8474,
+      "step": 193000
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.812433294901678e-05,
-      "loss": 2.7255,
-      "step": 178600
+      "learning_rate": 1.80558220787041e-05,
+      "loss": 2.8677,
+      "step": 193100
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8106485431071102e-05,
-      "loss": 2.7261,
-      "step": 178700
+      "learning_rate": 1.803927926258743e-05,
+      "loss": 2.8644,
+      "step": 193200
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8088637913125423e-05,
-      "loss": 2.7218,
-      "step": 178800
+      "learning_rate": 1.8022736446470756e-05,
+      "loss": 2.8638,
+      "step": 193300
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8070790395179744e-05,
-      "loss": 2.7325,
-      "step": 178900
+      "learning_rate": 1.8006193630354084e-05,
+      "loss": 2.8716,
+      "step": 193400
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8052942877234062e-05,
-      "loss": 2.7252,
-      "step": 179000
+      "learning_rate": 1.7989650814237412e-05,
+      "loss": 2.8616,
+      "step": 193500
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8035095359288384e-05,
-      "loss": 2.7204,
-      "step": 179100
+      "learning_rate": 1.7973107998120736e-05,
+      "loss": 2.8512,
+      "step": 193600
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.8017247841342705e-05,
-      "loss": 2.7163,
-      "step": 179200
+      "learning_rate": 1.7956565182004064e-05,
+      "loss": 2.8649,
+      "step": 193700
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7999400323397026e-05,
-      "loss": 2.7288,
-      "step": 179300
+      "learning_rate": 1.7940022365887392e-05,
+      "loss": 2.8562,
+      "step": 193800
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7981552805451347e-05,
-      "loss": 2.7302,
-      "step": 179400
+      "learning_rate": 1.7923479549770717e-05,
+      "loss": 2.8506,
+      "step": 193900
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7963705287505665e-05,
-      "loss": 2.7141,
-      "step": 179500
+      "learning_rate": 1.7906936733654045e-05,
+      "loss": 2.8527,
+      "step": 194000
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7945857769559986e-05,
-      "loss": 2.7133,
-      "step": 179600
+      "learning_rate": 1.789039391753737e-05,
+      "loss": 2.8693,
+      "step": 194100
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7928010251614307e-05,
-      "loss": 2.7207,
-      "step": 179700
+      "learning_rate": 1.7873851101420697e-05,
+      "loss": 2.853,
+      "step": 194200
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.791016273366863e-05,
-      "loss": 2.7191,
-      "step": 179800
+      "learning_rate": 1.7857308285304025e-05,
+      "loss": 2.864,
+      "step": 194300
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.789231521572295e-05,
-      "loss": 2.7188,
-      "step": 179900
+      "learning_rate": 1.784076546918735e-05,
+      "loss": 2.8564,
+      "step": 194400
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.787446769777727e-05,
-      "loss": 2.7159,
-      "step": 180000
+      "learning_rate": 1.782422265307068e-05,
+      "loss": 2.8654,
+      "step": 194500
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7856620179831592e-05,
-      "loss": 2.7176,
-      "step": 180100
+      "learning_rate": 1.7807679836954005e-05,
+      "loss": 2.8604,
+      "step": 194600
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7838772661885913e-05,
-      "loss": 2.7271,
-      "step": 180200
+      "learning_rate": 1.779113702083733e-05,
+      "loss": 2.8704,
+      "step": 194700
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7820925143940235e-05,
-      "loss": 2.7298,
-      "step": 180300
+      "learning_rate": 1.777459420472066e-05,
+      "loss": 2.8583,
+      "step": 194800
     },
     {
       "epoch": 0.64,
-      "learning_rate": 1.7803077625994556e-05,
-      "loss": 2.7158,
-      "step": 180400
+      "learning_rate": 1.7758051388603986e-05,
+      "loss": 2.8548,
+      "step": 194900
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.7785230108048874e-05,
-      "loss": 2.733,
-      "step": 180500
+      "epoch": 0.65,
+      "learning_rate": 1.774150857248731e-05,
+      "loss": 2.8537,
+      "step": 195000
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.7767382590103195e-05,
-      "loss": 2.7242,
-      "step": 180600
+      "epoch": 0.65,
+      "learning_rate": 1.7724965756370642e-05,
+      "loss": 2.8747,
+      "step": 195100
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7749535072157516e-05,
-      "loss": 2.7177,
-      "step": 180700
+      "learning_rate": 1.7708422940253966e-05,
+      "loss": 2.8611,
+      "step": 195200
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7731687554211837e-05,
-      "loss": 2.7187,
-      "step": 180800
+      "learning_rate": 1.769188012413729e-05,
+      "loss": 2.8474,
+      "step": 195300
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.771384003626616e-05,
-      "loss": 2.7196,
-      "step": 180900
+      "learning_rate": 1.767533730802062e-05,
+      "loss": 2.8734,
+      "step": 195400
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.769599251832048e-05,
-      "loss": 2.7254,
-      "step": 181000
+      "learning_rate": 1.7658794491903947e-05,
+      "loss": 2.8576,
+      "step": 195500
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7678145000374798e-05,
-      "loss": 2.717,
-      "step": 181100
+      "learning_rate": 1.7642251675787275e-05,
+      "loss": 2.869,
+      "step": 195600
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.766029748242912e-05,
-      "loss": 2.7162,
-      "step": 181200
+      "learning_rate": 1.76257088596706e-05,
+      "loss": 2.8603,
+      "step": 195700
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.764244996448344e-05,
-      "loss": 2.7167,
-      "step": 181300
+      "learning_rate": 1.7609166043553927e-05,
+      "loss": 2.8682,
+      "step": 195800
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.762460244653776e-05,
-      "loss": 2.7216,
-      "step": 181400
+      "learning_rate": 1.7592623227437255e-05,
+      "loss": 2.8605,
+      "step": 195900
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7606754928592082e-05,
-      "loss": 2.7254,
-      "step": 181500
+      "learning_rate": 1.757608041132058e-05,
+      "loss": 2.8511,
+      "step": 196000
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.75889074106464e-05,
-      "loss": 2.72,
-      "step": 181600
+      "learning_rate": 1.7559537595203907e-05,
+      "loss": 2.8575,
+      "step": 196100
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.757105989270072e-05,
-      "loss": 2.7132,
-      "step": 181700
+      "learning_rate": 1.7542994779087235e-05,
+      "loss": 2.8677,
+      "step": 196200
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7553212374755043e-05,
-      "loss": 2.723,
-      "step": 181800
+      "learning_rate": 1.752645196297056e-05,
+      "loss": 2.8633,
+      "step": 196300
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7535364856809364e-05,
-      "loss": 2.7191,
-      "step": 181900
+      "learning_rate": 1.7509909146853888e-05,
+      "loss": 2.8526,
+      "step": 196400
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7517517338863685e-05,
-      "loss": 2.7216,
-      "step": 182000
+      "learning_rate": 1.7493366330737216e-05,
+      "loss": 2.8704,
+      "step": 196500
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7499669820918006e-05,
-      "loss": 2.7184,
-      "step": 182100
+      "learning_rate": 1.747682351462054e-05,
+      "loss": 2.857,
+      "step": 196600
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7481822302972324e-05,
-      "loss": 2.7303,
-      "step": 182200
+      "learning_rate": 1.7460280698503868e-05,
+      "loss": 2.8704,
+      "step": 196700
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7463974785026645e-05,
-      "loss": 2.7074,
-      "step": 182300
+      "learning_rate": 1.7443737882387196e-05,
+      "loss": 2.8536,
+      "step": 196800
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7446127267080967e-05,
-      "loss": 2.7286,
-      "step": 182400
+      "learning_rate": 1.742719506627052e-05,
+      "loss": 2.8515,
+      "step": 196900
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7428279749135288e-05,
-      "loss": 2.7204,
-      "step": 182500
+      "learning_rate": 1.741065225015385e-05,
+      "loss": 2.8552,
+      "step": 197000
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.741043223118961e-05,
-      "loss": 2.7159,
-      "step": 182600
+      "learning_rate": 1.7394109434037176e-05,
+      "loss": 2.8656,
+      "step": 197100
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.739258471324393e-05,
-      "loss": 2.7101,
-      "step": 182700
+      "learning_rate": 1.73775666179205e-05,
+      "loss": 2.8709,
+      "step": 197200
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.737473719529825e-05,
-      "loss": 2.7151,
-      "step": 182800
+      "learning_rate": 1.736102380180383e-05,
+      "loss": 2.853,
+      "step": 197300
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7356889677352573e-05,
-      "loss": 2.7056,
-      "step": 182900
+      "learning_rate": 1.7344480985687157e-05,
+      "loss": 2.862,
+      "step": 197400
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7339042159406894e-05,
-      "loss": 2.7251,
-      "step": 183000
+      "learning_rate": 1.732793816957048e-05,
+      "loss": 2.8684,
+      "step": 197500
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7321194641461215e-05,
-      "loss": 2.7264,
-      "step": 183100
+      "learning_rate": 1.731139535345381e-05,
+      "loss": 2.8601,
+      "step": 197600
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7303347123515533e-05,
-      "loss": 2.7249,
-      "step": 183200
+      "learning_rate": 1.7294852537337137e-05,
+      "loss": 2.8411,
+      "step": 197700
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7285499605569854e-05,
-      "loss": 2.7095,
-      "step": 183300
+      "learning_rate": 1.7278309721220465e-05,
+      "loss": 2.8567,
+      "step": 197800
     },
     {
       "epoch": 0.65,
-      "learning_rate": 1.7267652087624175e-05,
-      "loss": 2.7235,
-      "step": 183400
+      "learning_rate": 1.726176690510379e-05,
+      "loss": 2.8572,
+      "step": 197900
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7249804569678497e-05,
-      "loss": 2.7213,
-      "step": 183500
+      "learning_rate": 1.7245224088987118e-05,
+      "loss": 2.8632,
+      "step": 198000
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7231957051732818e-05,
-      "loss": 2.7328,
-      "step": 183600
+      "learning_rate": 1.7228681272870446e-05,
+      "loss": 2.8686,
+      "step": 198100
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7214109533787136e-05,
-      "loss": 2.7092,
-      "step": 183700
+      "learning_rate": 1.721213845675377e-05,
+      "loss": 2.8528,
+      "step": 198200
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7196262015841457e-05,
-      "loss": 2.7146,
-      "step": 183800
+      "learning_rate": 1.7195595640637098e-05,
+      "loss": 2.8469,
+      "step": 198300
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7178414497895778e-05,
-      "loss": 2.7208,
-      "step": 183900
+      "learning_rate": 1.7179052824520426e-05,
+      "loss": 2.8626,
+      "step": 198400
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.71605669799501e-05,
-      "loss": 2.723,
-      "step": 184000
+      "learning_rate": 1.716251000840375e-05,
+      "loss": 2.8387,
+      "step": 198500
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.714271946200442e-05,
-      "loss": 2.7104,
-      "step": 184100
+      "learning_rate": 1.714596719228708e-05,
+      "loss": 2.8597,
+      "step": 198600
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7124871944058742e-05,
-      "loss": 2.7221,
-      "step": 184200
+      "learning_rate": 1.7129424376170406e-05,
+      "loss": 2.8744,
+      "step": 198700
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.710702442611306e-05,
-      "loss": 2.7253,
-      "step": 184300
+      "learning_rate": 1.711288156005373e-05,
+      "loss": 2.8657,
+      "step": 198800
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.708917690816738e-05,
-      "loss": 2.712,
-      "step": 184400
+      "learning_rate": 1.709633874393706e-05,
+      "loss": 2.8655,
+      "step": 198900
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7071329390221702e-05,
-      "loss": 2.723,
-      "step": 184500
+      "learning_rate": 1.7079795927820387e-05,
+      "loss": 2.8516,
+      "step": 199000
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7053481872276023e-05,
-      "loss": 2.7147,
-      "step": 184600
+      "learning_rate": 1.706325311170371e-05,
+      "loss": 2.8494,
+      "step": 199100
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7035634354330344e-05,
-      "loss": 2.7115,
-      "step": 184700
+      "learning_rate": 1.704671029558704e-05,
+      "loss": 2.8534,
+      "step": 199200
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.7017786836384662e-05,
-      "loss": 2.7187,
-      "step": 184800
+      "learning_rate": 1.7030167479470367e-05,
+      "loss": 2.8589,
+      "step": 199300
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6999939318438983e-05,
-      "loss": 2.7232,
-      "step": 184900
+      "learning_rate": 1.701362466335369e-05,
+      "loss": 2.8537,
+      "step": 199400
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6982091800493305e-05,
-      "loss": 2.7248,
-      "step": 185000
+      "learning_rate": 1.699708184723702e-05,
+      "loss": 2.8602,
+      "step": 199500
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6964244282547626e-05,
-      "loss": 2.7092,
-      "step": 185100
+      "learning_rate": 1.6980539031120347e-05,
+      "loss": 2.8622,
+      "step": 199600
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.694639676460195e-05,
-      "loss": 2.7228,
-      "step": 185200
+      "learning_rate": 1.6963996215003675e-05,
+      "loss": 2.8481,
+      "step": 199700
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.692854924665627e-05,
-      "loss": 2.7268,
-      "step": 185300
+      "learning_rate": 1.6947453398887e-05,
+      "loss": 2.8486,
+      "step": 199800
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.691070172871059e-05,
-      "loss": 2.7142,
-      "step": 185400
+      "learning_rate": 1.6930910582770328e-05,
+      "loss": 2.8564,
+      "step": 199900
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.689285421076491e-05,
-      "loss": 2.7131,
-      "step": 185500
+      "learning_rate": 1.6914367766653656e-05,
+      "loss": 2.853,
+      "step": 200000
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6875006692819232e-05,
-      "loss": 2.7241,
-      "step": 185600
+      "learning_rate": 1.689782495053698e-05,
+      "loss": 2.8697,
+      "step": 200100
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6857159174873553e-05,
-      "loss": 2.7147,
-      "step": 185700
+      "learning_rate": 1.6881282134420308e-05,
+      "loss": 2.8682,
+      "step": 200200
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.683931165692787e-05,
-      "loss": 2.7145,
-      "step": 185800
+      "learning_rate": 1.6864739318303636e-05,
+      "loss": 2.8657,
+      "step": 200300
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6821464138982192e-05,
-      "loss": 2.7177,
-      "step": 185900
+      "learning_rate": 1.684819650218696e-05,
+      "loss": 2.8491,
+      "step": 200400
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6803616621036513e-05,
-      "loss": 2.7221,
-      "step": 186000
+      "learning_rate": 1.6831653686070285e-05,
+      "loss": 2.8492,
+      "step": 200500
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6785769103090835e-05,
-      "loss": 2.7218,
-      "step": 186100
+      "learning_rate": 1.6815110869953616e-05,
+      "loss": 2.8622,
+      "step": 200600
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6767921585145156e-05,
-      "loss": 2.7185,
-      "step": 186200
+      "learning_rate": 1.679856805383694e-05,
+      "loss": 2.8634,
+      "step": 200700
     },
     {
       "epoch": 0.66,
-      "learning_rate": 1.6750074067199477e-05,
-      "loss": 2.7084,
-      "step": 186300
+      "learning_rate": 1.678202523772027e-05,
+      "loss": 2.8588,
+      "step": 200800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6765482421603597e-05,
+      "loss": 2.8649,
+      "step": 200900
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6732226549253795e-05,
-      "loss": 2.7109,
-      "step": 186400
+      "learning_rate": 1.674893960548692e-05,
+      "loss": 2.8671,
+      "step": 201000
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6714379031308116e-05,
-      "loss": 2.7236,
-      "step": 186500
+      "learning_rate": 1.673239678937025e-05,
+      "loss": 2.8497,
+      "step": 201100
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6696531513362437e-05,
-      "loss": 2.732,
-      "step": 186600
+      "learning_rate": 1.6715853973253577e-05,
+      "loss": 2.8542,
+      "step": 201200
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.667868399541676e-05,
-      "loss": 2.7222,
-      "step": 186700
+      "learning_rate": 1.6699311157136902e-05,
+      "loss": 2.8554,
+      "step": 201300
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.666083647747108e-05,
-      "loss": 2.7112,
-      "step": 186800
+      "learning_rate": 1.668276834102023e-05,
+      "loss": 2.8534,
+      "step": 201400
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6642988959525398e-05,
-      "loss": 2.7275,
-      "step": 186900
+      "learning_rate": 1.6666225524903558e-05,
+      "loss": 2.8591,
+      "step": 201500
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.662514144157972e-05,
-      "loss": 2.7227,
-      "step": 187000
+      "learning_rate": 1.6649682708786882e-05,
+      "loss": 2.8407,
+      "step": 201600
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.660729392363404e-05,
-      "loss": 2.723,
-      "step": 187100
+      "learning_rate": 1.663313989267021e-05,
+      "loss": 2.8678,
+      "step": 201700
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.658944640568836e-05,
-      "loss": 2.732,
-      "step": 187200
+      "learning_rate": 1.6616597076553535e-05,
+      "loss": 2.8625,
+      "step": 201800
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6571598887742682e-05,
-      "loss": 2.7154,
-      "step": 187300
+      "learning_rate": 1.6600054260436866e-05,
+      "loss": 2.8606,
+      "step": 201900
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6553751369797004e-05,
-      "loss": 2.7151,
-      "step": 187400
+      "learning_rate": 1.658351144432019e-05,
+      "loss": 2.8665,
+      "step": 202000
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.653590385185132e-05,
-      "loss": 2.7072,
-      "step": 187500
+      "learning_rate": 1.6566968628203515e-05,
+      "loss": 2.8668,
+      "step": 202100
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6518056333905643e-05,
-      "loss": 2.7216,
-      "step": 187600
+      "learning_rate": 1.6550425812086846e-05,
+      "loss": 2.8491,
+      "step": 202200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.653388299597017e-05,
+      "loss": 2.8694,
+      "step": 202300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6517340179853495e-05,
+      "loss": 2.8619,
+      "step": 202400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6500797363736827e-05,
+      "loss": 2.8665,
+      "step": 202500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.648425454762015e-05,
+      "loss": 2.8674,
+      "step": 202600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6467711731503476e-05,
+      "loss": 2.855,
+      "step": 202700
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6500208815959967e-05,
-      "loss": 2.7015,
-      "step": 187700
+      "learning_rate": 1.6451168915386807e-05,
+      "loss": 2.868,
+      "step": 202800
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.648236129801429e-05,
-      "loss": 2.7158,
-      "step": 187800
+      "learning_rate": 1.643462609927013e-05,
+      "loss": 2.8581,
+      "step": 202900
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6464513780068606e-05,
-      "loss": 2.7235,
-      "step": 187900
+      "learning_rate": 1.641808328315346e-05,
+      "loss": 2.8631,
+      "step": 203000
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6446666262122928e-05,
-      "loss": 2.7088,
-      "step": 188000
+      "learning_rate": 1.6401540467036784e-05,
+      "loss": 2.8618,
+      "step": 203100
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.642881874417725e-05,
-      "loss": 2.7184,
-      "step": 188100
+      "learning_rate": 1.6384997650920112e-05,
+      "loss": 2.8658,
+      "step": 203200
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.641097122623157e-05,
-      "loss": 2.7156,
-      "step": 188200
+      "learning_rate": 1.636845483480344e-05,
+      "loss": 2.8455,
+      "step": 203300
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.639312370828589e-05,
-      "loss": 2.7184,
-      "step": 188300
+      "learning_rate": 1.6351912018686764e-05,
+      "loss": 2.8525,
+      "step": 203400
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6375276190340212e-05,
-      "loss": 2.7243,
-      "step": 188400
+      "learning_rate": 1.6335369202570092e-05,
+      "loss": 2.868,
+      "step": 203500
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.635742867239453e-05,
-      "loss": 2.7248,
-      "step": 188500
+      "learning_rate": 1.631882638645342e-05,
+      "loss": 2.8555,
+      "step": 203600
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.633958115444885e-05,
-      "loss": 2.7154,
-      "step": 188600
+      "learning_rate": 1.6302283570336745e-05,
+      "loss": 2.8556,
+      "step": 203700
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6321733636503173e-05,
-      "loss": 2.7176,
-      "step": 188700
+      "learning_rate": 1.6285740754220073e-05,
+      "loss": 2.8521,
+      "step": 203800
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6303886118557494e-05,
-      "loss": 2.7134,
-      "step": 188800
+      "learning_rate": 1.62691979381034e-05,
+      "loss": 2.8492,
+      "step": 203900
     },
     {
       "epoch": 0.67,
-      "learning_rate": 1.6286038600611815e-05,
-      "loss": 2.7259,
-      "step": 188900
+      "learning_rate": 1.6252655121986725e-05,
+      "loss": 2.8548,
+      "step": 204000
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.6268191082666133e-05,
-      "loss": 2.7226,
-      "step": 189000
+      "epoch": 0.68,
+      "learning_rate": 1.6236112305870056e-05,
+      "loss": 2.8539,
+      "step": 204100
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.6250343564720454e-05,
-      "loss": 2.7062,
-      "step": 189100
+      "epoch": 0.68,
+      "learning_rate": 1.621956948975338e-05,
+      "loss": 2.8445,
+      "step": 204200
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6232496046774775e-05,
-      "loss": 2.7082,
-      "step": 189200
+      "learning_rate": 1.6203026673636706e-05,
+      "loss": 2.8369,
+      "step": 204300
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6214648528829097e-05,
-      "loss": 2.712,
-      "step": 189300
+      "learning_rate": 1.6186483857520033e-05,
+      "loss": 2.8598,
+      "step": 204400
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6196801010883418e-05,
-      "loss": 2.7137,
-      "step": 189400
+      "learning_rate": 1.616994104140336e-05,
+      "loss": 2.847,
+      "step": 204500
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6178953492937736e-05,
-      "loss": 2.7242,
-      "step": 189500
+      "learning_rate": 1.6153398225286686e-05,
+      "loss": 2.8633,
+      "step": 204600
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6161105974992057e-05,
-      "loss": 2.7074,
-      "step": 189600
+      "learning_rate": 1.6136855409170014e-05,
+      "loss": 2.8542,
+      "step": 204700
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6143258457046378e-05,
-      "loss": 2.7128,
-      "step": 189700
+      "learning_rate": 1.6120312593053342e-05,
+      "loss": 2.8586,
+      "step": 204800
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.61254109391007e-05,
-      "loss": 2.726,
-      "step": 189800
+      "learning_rate": 1.6103769776936666e-05,
+      "loss": 2.8573,
+      "step": 204900
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.610756342115502e-05,
-      "loss": 2.7125,
-      "step": 189900
+      "learning_rate": 1.6087226960819994e-05,
+      "loss": 2.8315,
+      "step": 205000
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6089715903209342e-05,
-      "loss": 2.7131,
-      "step": 190000
+      "learning_rate": 1.6070684144703322e-05,
+      "loss": 2.8482,
+      "step": 205100
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.607186838526366e-05,
-      "loss": 2.7301,
-      "step": 190100
+      "learning_rate": 1.605414132858665e-05,
+      "loss": 2.8812,
+      "step": 205200
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.605402086731798e-05,
-      "loss": 2.7243,
-      "step": 190200
+      "learning_rate": 1.6037598512469975e-05,
+      "loss": 2.855,
+      "step": 205300
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6036173349372305e-05,
-      "loss": 2.715,
-      "step": 190300
+      "learning_rate": 1.6021055696353303e-05,
+      "loss": 2.862,
+      "step": 205400
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6018325831426627e-05,
-      "loss": 2.7066,
-      "step": 190400
+      "learning_rate": 1.600451288023663e-05,
+      "loss": 2.8527,
+      "step": 205500
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.6000478313480944e-05,
-      "loss": 2.7182,
-      "step": 190500
+      "learning_rate": 1.5987970064119955e-05,
+      "loss": 2.85,
+      "step": 205600
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5982630795535266e-05,
-      "loss": 2.7147,
-      "step": 190600
+      "learning_rate": 1.5971427248003283e-05,
+      "loss": 2.8581,
+      "step": 205700
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5964783277589587e-05,
-      "loss": 2.7144,
-      "step": 190700
+      "learning_rate": 1.595488443188661e-05,
+      "loss": 2.8535,
+      "step": 205800
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5946935759643908e-05,
-      "loss": 2.7222,
-      "step": 190800
+      "learning_rate": 1.5938341615769935e-05,
+      "loss": 2.8582,
+      "step": 205900
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.592908824169823e-05,
-      "loss": 2.7202,
-      "step": 190900
+      "learning_rate": 1.5921798799653263e-05,
+      "loss": 2.8497,
+      "step": 206000
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.591124072375255e-05,
-      "loss": 2.7086,
-      "step": 191000
+      "learning_rate": 1.590525598353659e-05,
+      "loss": 2.8544,
+      "step": 206100
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.589339320580687e-05,
-      "loss": 2.7141,
-      "step": 191100
+      "learning_rate": 1.5888713167419916e-05,
+      "loss": 2.851,
+      "step": 206200
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.587554568786119e-05,
-      "loss": 2.7148,
-      "step": 191200
+      "learning_rate": 1.5872170351303244e-05,
+      "loss": 2.8406,
+      "step": 206300
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.585769816991551e-05,
-      "loss": 2.7037,
-      "step": 191300
+      "learning_rate": 1.585562753518657e-05,
+      "loss": 2.8449,
+      "step": 206400
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5839850651969832e-05,
-      "loss": 2.7298,
-      "step": 191400
+      "learning_rate": 1.5839084719069896e-05,
+      "loss": 2.8653,
+      "step": 206500
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5822003134024153e-05,
-      "loss": 2.7071,
-      "step": 191500
+      "learning_rate": 1.5822541902953224e-05,
+      "loss": 2.845,
+      "step": 206600
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.580415561607847e-05,
-      "loss": 2.71,
-      "step": 191600
+      "learning_rate": 1.5805999086836552e-05,
+      "loss": 2.8482,
+      "step": 206700
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5786308098132792e-05,
-      "loss": 2.7058,
-      "step": 191700
+      "learning_rate": 1.5789456270719877e-05,
+      "loss": 2.8358,
+      "step": 206800
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5768460580187113e-05,
-      "loss": 2.7256,
-      "step": 191800
+      "learning_rate": 1.5772913454603204e-05,
+      "loss": 2.8393,
+      "step": 206900
     },
     {
       "epoch": 0.68,
-      "learning_rate": 1.5750613062241435e-05,
-      "loss": 2.7194,
-      "step": 191900
+      "learning_rate": 1.5756370638486532e-05,
+      "loss": 2.8469,
+      "step": 207000
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5732765544295756e-05,
-      "loss": 2.7117,
-      "step": 192000
+      "learning_rate": 1.573982782236986e-05,
+      "loss": 2.8568,
+      "step": 207100
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5714918026350077e-05,
-      "loss": 2.7162,
-      "step": 192100
+      "learning_rate": 1.5723285006253185e-05,
+      "loss": 2.8597,
+      "step": 207200
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5697070508404395e-05,
-      "loss": 2.7219,
-      "step": 192200
+      "learning_rate": 1.5706742190136513e-05,
+      "loss": 2.8483,
+      "step": 207300
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5679222990458716e-05,
-      "loss": 2.716,
-      "step": 192300
+      "learning_rate": 1.569019937401984e-05,
+      "loss": 2.86,
+      "step": 207400
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5661375472513037e-05,
-      "loss": 2.7156,
-      "step": 192400
+      "learning_rate": 1.5673656557903165e-05,
+      "loss": 2.8662,
+      "step": 207500
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.564352795456736e-05,
-      "loss": 2.7103,
-      "step": 192500
+      "learning_rate": 1.5657113741786493e-05,
+      "loss": 2.854,
+      "step": 207600
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.562568043662168e-05,
-      "loss": 2.7085,
-      "step": 192600
+      "learning_rate": 1.564057092566982e-05,
+      "loss": 2.8662,
+      "step": 207700
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5607832918675998e-05,
-      "loss": 2.7151,
-      "step": 192700
+      "learning_rate": 1.5624028109553146e-05,
+      "loss": 2.8529,
+      "step": 207800
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5589985400730322e-05,
-      "loss": 2.711,
-      "step": 192800
+      "learning_rate": 1.560748529343647e-05,
+      "loss": 2.8573,
+      "step": 207900
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5572137882784643e-05,
-      "loss": 2.7089,
-      "step": 192900
+      "learning_rate": 1.55909424773198e-05,
+      "loss": 2.8514,
+      "step": 208000
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5554290364838965e-05,
-      "loss": 2.7197,
-      "step": 193000
+      "learning_rate": 1.5574399661203126e-05,
+      "loss": 2.8628,
+      "step": 208100
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5536442846893286e-05,
-      "loss": 2.7035,
-      "step": 193100
+      "learning_rate": 1.5557856845086454e-05,
+      "loss": 2.8512,
+      "step": 208200
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5518595328947604e-05,
-      "loss": 2.724,
-      "step": 193200
+      "learning_rate": 1.5541314028969782e-05,
+      "loss": 2.8553,
+      "step": 208300
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5500747811001925e-05,
-      "loss": 2.721,
-      "step": 193300
+      "learning_rate": 1.5524771212853106e-05,
+      "loss": 2.8566,
+      "step": 208400
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5482900293056246e-05,
-      "loss": 2.7175,
-      "step": 193400
+      "learning_rate": 1.5508228396736434e-05,
+      "loss": 2.8597,
+      "step": 208500
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5465052775110567e-05,
-      "loss": 2.7098,
-      "step": 193500
+      "learning_rate": 1.5491685580619762e-05,
+      "loss": 2.8598,
+      "step": 208600
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.544720525716489e-05,
-      "loss": 2.7112,
-      "step": 193600
+      "learning_rate": 1.5475142764503087e-05,
+      "loss": 2.847,
+      "step": 208700
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5429357739219206e-05,
-      "loss": 2.7137,
-      "step": 193700
+      "learning_rate": 1.5458599948386415e-05,
+      "loss": 2.8439,
+      "step": 208800
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5411510221273528e-05,
-      "loss": 2.7177,
-      "step": 193800
+      "learning_rate": 1.5442057132269743e-05,
+      "loss": 2.8749,
+      "step": 208900
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.539366270332785e-05,
-      "loss": 2.7109,
-      "step": 193900
+      "learning_rate": 1.5425514316153067e-05,
+      "loss": 2.8684,
+      "step": 209000
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.537581518538217e-05,
-      "loss": 2.7027,
-      "step": 194000
+      "learning_rate": 1.5408971500036395e-05,
+      "loss": 2.861,
+      "step": 209100
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.535796766743649e-05,
-      "loss": 2.7139,
-      "step": 194100
+      "learning_rate": 1.539242868391972e-05,
+      "loss": 2.8457,
+      "step": 209200
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5340120149490812e-05,
-      "loss": 2.7126,
-      "step": 194200
+      "learning_rate": 1.537588586780305e-05,
+      "loss": 2.8469,
+      "step": 209300
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.532227263154513e-05,
-      "loss": 2.7025,
-      "step": 194300
+      "learning_rate": 1.5359343051686375e-05,
+      "loss": 2.857,
+      "step": 209400
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.530442511359945e-05,
-      "loss": 2.7097,
-      "step": 194400
+      "learning_rate": 1.53428002355697e-05,
+      "loss": 2.8544,
+      "step": 209500
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5286577595653773e-05,
-      "loss": 2.7117,
-      "step": 194500
+      "learning_rate": 1.532625741945303e-05,
+      "loss": 2.8464,
+      "step": 209600
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5268730077708094e-05,
-      "loss": 2.7228,
-      "step": 194600
+      "learning_rate": 1.5309714603336356e-05,
+      "loss": 2.8386,
+      "step": 209700
     },
     {
       "epoch": 0.69,
-      "learning_rate": 1.5250882559762413e-05,
-      "loss": 2.7108,
-      "step": 194700
+      "learning_rate": 1.529317178721968e-05,
+      "loss": 2.853,
+      "step": 209800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527662897110301e-05,
+      "loss": 2.8418,
+      "step": 209900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5260086154986336e-05,
+      "loss": 2.8395,
+      "step": 210000
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5233035041816735e-05,
-      "loss": 2.6974,
-      "step": 194800
+      "learning_rate": 1.5243543338869662e-05,
+      "loss": 2.8516,
+      "step": 210100
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5215187523871056e-05,
-      "loss": 2.7131,
-      "step": 194900
+      "learning_rate": 1.522700052275299e-05,
+      "loss": 2.8483,
+      "step": 210200
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5197340005925375e-05,
-      "loss": 2.7097,
-      "step": 195000
+      "learning_rate": 1.5210457706636317e-05,
+      "loss": 2.852,
+      "step": 210300
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5179492487979697e-05,
-      "loss": 2.7097,
-      "step": 195100
+      "learning_rate": 1.5193914890519644e-05,
+      "loss": 2.842,
+      "step": 210400
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5161644970034016e-05,
-      "loss": 2.7182,
-      "step": 195200
+      "learning_rate": 1.517737207440297e-05,
+      "loss": 2.8542,
+      "step": 210500
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5143797452088337e-05,
-      "loss": 2.7113,
-      "step": 195300
+      "learning_rate": 1.5160829258286297e-05,
+      "loss": 2.8528,
+      "step": 210600
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.512594993414266e-05,
-      "loss": 2.7196,
-      "step": 195400
+      "learning_rate": 1.5144286442169625e-05,
+      "loss": 2.8462,
+      "step": 210700
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5108102416196981e-05,
-      "loss": 2.7211,
-      "step": 195500
+      "learning_rate": 1.5127743626052951e-05,
+      "loss": 2.8487,
+      "step": 210800
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5090254898251303e-05,
-      "loss": 2.7129,
-      "step": 195600
+      "learning_rate": 1.5111200809936277e-05,
+      "loss": 2.8459,
+      "step": 210900
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5072407380305622e-05,
-      "loss": 2.7124,
-      "step": 195700
+      "learning_rate": 1.5094657993819605e-05,
+      "loss": 2.8386,
+      "step": 211000
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5054559862359943e-05,
-      "loss": 2.7142,
-      "step": 195800
+      "learning_rate": 1.5078115177702931e-05,
+      "loss": 2.8587,
+      "step": 211100
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5036712344414265e-05,
-      "loss": 2.71,
-      "step": 195900
+      "learning_rate": 1.5061572361586258e-05,
+      "loss": 2.8425,
+      "step": 211200
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5018864826468584e-05,
-      "loss": 2.7202,
-      "step": 196000
+      "learning_rate": 1.5045029545469586e-05,
+      "loss": 2.8554,
+      "step": 211300
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.5001017308522905e-05,
-      "loss": 2.7147,
-      "step": 196100
+      "learning_rate": 1.5028486729352912e-05,
+      "loss": 2.8483,
+      "step": 211400
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4983169790577225e-05,
-      "loss": 2.705,
-      "step": 196200
+      "learning_rate": 1.501194391323624e-05,
+      "loss": 2.8396,
+      "step": 211500
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4965322272631546e-05,
-      "loss": 2.7123,
-      "step": 196300
+      "learning_rate": 1.4995401097119566e-05,
+      "loss": 2.8483,
+      "step": 211600
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4947474754685867e-05,
-      "loss": 2.7156,
-      "step": 196400
+      "learning_rate": 1.4978858281002892e-05,
+      "loss": 2.8637,
+      "step": 211700
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4929627236740187e-05,
-      "loss": 2.716,
-      "step": 196500
+      "learning_rate": 1.496231546488622e-05,
+      "loss": 2.8611,
+      "step": 211800
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4911779718794508e-05,
-      "loss": 2.7234,
-      "step": 196600
+      "learning_rate": 1.4945772648769546e-05,
+      "loss": 2.8586,
+      "step": 211900
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.489393220084883e-05,
-      "loss": 2.7084,
-      "step": 196700
+      "learning_rate": 1.4929229832652873e-05,
+      "loss": 2.8486,
+      "step": 212000
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4876084682903149e-05,
-      "loss": 2.7275,
-      "step": 196800
+      "learning_rate": 1.49126870165362e-05,
+      "loss": 2.8633,
+      "step": 212100
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.485823716495747e-05,
-      "loss": 2.7241,
-      "step": 196900
+      "learning_rate": 1.4896144200419527e-05,
+      "loss": 2.8639,
+      "step": 212200
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4840389647011791e-05,
-      "loss": 2.7109,
-      "step": 197000
+      "learning_rate": 1.4879601384302851e-05,
+      "loss": 2.8584,
+      "step": 212300
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.482254212906611e-05,
-      "loss": 2.7211,
-      "step": 197100
+      "learning_rate": 1.4863058568186181e-05,
+      "loss": 2.8558,
+      "step": 212400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4846515752069507e-05,
+      "loss": 2.8408,
+      "step": 212500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4829972935952835e-05,
+      "loss": 2.8569,
+      "step": 212600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4813430119836161e-05,
+      "loss": 2.8612,
+      "step": 212700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4796887303719486e-05,
+      "loss": 2.8472,
+      "step": 212800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4780344487602815e-05,
+      "loss": 2.8533,
+      "step": 212900
     },
     {
       "epoch": 0.7,
-      "learning_rate": 1.4804694611120432e-05,
-      "loss": 2.717,
-      "step": 197200
+      "learning_rate": 1.4763801671486142e-05,
+      "loss": 2.8575,
+      "step": 213000
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.4786847093174751e-05,
-      "loss": 2.7164,
-      "step": 197300
+      "epoch": 0.71,
+      "learning_rate": 1.4747258855369466e-05,
+      "loss": 2.8472,
+      "step": 213100
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.4768999575229073e-05,
-      "loss": 2.7071,
-      "step": 197400
+      "epoch": 0.71,
+      "learning_rate": 1.4730716039252796e-05,
+      "loss": 2.841,
+      "step": 213200
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.4751152057283394e-05,
-      "loss": 2.7143,
-      "step": 197500
+      "epoch": 0.71,
+      "learning_rate": 1.4714173223136122e-05,
+      "loss": 2.8529,
+      "step": 213300
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4733304539337713e-05,
-      "loss": 2.7101,
-      "step": 197600
+      "learning_rate": 1.469763040701945e-05,
+      "loss": 2.8317,
+      "step": 213400
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4715457021392035e-05,
-      "loss": 2.703,
-      "step": 197700
+      "learning_rate": 1.4681087590902776e-05,
+      "loss": 2.8481,
+      "step": 213500
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4697609503446356e-05,
-      "loss": 2.7191,
-      "step": 197800
+      "learning_rate": 1.46645447747861e-05,
+      "loss": 2.8639,
+      "step": 213600
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4679761985500679e-05,
-      "loss": 2.6988,
-      "step": 197900
+      "learning_rate": 1.464800195866943e-05,
+      "loss": 2.8554,
+      "step": 213700
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4661914467554998e-05,
-      "loss": 2.7139,
-      "step": 198000
+      "learning_rate": 1.4631459142552757e-05,
+      "loss": 2.8518,
+      "step": 213800
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.464406694960932e-05,
-      "loss": 2.7126,
-      "step": 198100
+      "learning_rate": 1.4614916326436081e-05,
+      "loss": 2.862,
+      "step": 213900
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.462621943166364e-05,
-      "loss": 2.7098,
-      "step": 198200
+      "learning_rate": 1.459837351031941e-05,
+      "loss": 2.8572,
+      "step": 214000
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.460837191371796e-05,
-      "loss": 2.7169,
-      "step": 198300
+      "learning_rate": 1.4581830694202735e-05,
+      "loss": 2.8544,
+      "step": 214100
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4590524395772281e-05,
-      "loss": 2.7155,
-      "step": 198400
+      "learning_rate": 1.4565287878086061e-05,
+      "loss": 2.8504,
+      "step": 214200
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4572676877826603e-05,
-      "loss": 2.7074,
-      "step": 198500
+      "learning_rate": 1.4548745061969391e-05,
+      "loss": 2.8523,
+      "step": 214300
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4554829359880922e-05,
-      "loss": 2.7125,
-      "step": 198600
+      "learning_rate": 1.4532202245852716e-05,
+      "loss": 2.846,
+      "step": 214400
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4536981841935243e-05,
-      "loss": 2.7075,
-      "step": 198700
+      "learning_rate": 1.4515659429736045e-05,
+      "loss": 2.8558,
+      "step": 214500
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4519134323989565e-05,
-      "loss": 2.7093,
-      "step": 198800
+      "learning_rate": 1.4499116613619371e-05,
+      "loss": 2.848,
+      "step": 214600
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4501286806043884e-05,
-      "loss": 2.723,
-      "step": 198900
+      "learning_rate": 1.4482573797502696e-05,
+      "loss": 2.8652,
+      "step": 214700
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4483439288098205e-05,
-      "loss": 2.7153,
-      "step": 199000
+      "learning_rate": 1.4466030981386026e-05,
+      "loss": 2.8552,
+      "step": 214800
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4465591770152525e-05,
-      "loss": 2.722,
-      "step": 199100
+      "learning_rate": 1.444948816526935e-05,
+      "loss": 2.8428,
+      "step": 214900
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4447744252206846e-05,
-      "loss": 2.7141,
-      "step": 199200
+      "learning_rate": 1.4432945349152676e-05,
+      "loss": 2.8392,
+      "step": 215000
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4429896734261167e-05,
-      "loss": 2.7116,
-      "step": 199300
+      "learning_rate": 1.4416402533036006e-05,
+      "loss": 2.8477,
+      "step": 215100
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4412049216315487e-05,
-      "loss": 2.7091,
-      "step": 199400
+      "learning_rate": 1.439985971691933e-05,
+      "loss": 2.8547,
+      "step": 215200
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4394201698369808e-05,
-      "loss": 2.7091,
-      "step": 199500
+      "learning_rate": 1.4383316900802657e-05,
+      "loss": 2.8501,
+      "step": 215300
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.437635418042413e-05,
-      "loss": 2.7183,
-      "step": 199600
+      "learning_rate": 1.4366774084685985e-05,
+      "loss": 2.846,
+      "step": 215400
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4358506662478449e-05,
-      "loss": 2.7198,
-      "step": 199700
+      "learning_rate": 1.4350231268569311e-05,
+      "loss": 2.8413,
+      "step": 215500
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.434065914453277e-05,
-      "loss": 2.7075,
-      "step": 199800
+      "learning_rate": 1.433368845245264e-05,
+      "loss": 2.8521,
+      "step": 215600
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4322811626587091e-05,
-      "loss": 2.7077,
-      "step": 199900
+      "learning_rate": 1.4317145636335965e-05,
+      "loss": 2.8584,
+      "step": 215700
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.430496410864141e-05,
-      "loss": 2.7082,
-      "step": 200000
+      "learning_rate": 1.4300602820219291e-05,
+      "loss": 2.8485,
+      "step": 215800
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4287116590695732e-05,
-      "loss": 2.7021,
-      "step": 200100
+      "learning_rate": 1.4284060004102621e-05,
+      "loss": 2.8546,
+      "step": 215900
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4269269072750051e-05,
-      "loss": 2.7069,
-      "step": 200200
+      "learning_rate": 1.4267517187985945e-05,
+      "loss": 2.8527,
+      "step": 216000
     },
     {
       "epoch": 0.71,
-      "learning_rate": 1.4251421554804373e-05,
-      "loss": 2.7115,
-      "step": 200300
+      "learning_rate": 1.4250974371869272e-05,
+      "loss": 2.8563,
+      "step": 216100
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4233574036858694e-05,
-      "loss": 2.721,
-      "step": 200400
+      "learning_rate": 1.42344315557526e-05,
+      "loss": 2.8541,
+      "step": 216200
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4215726518913017e-05,
-      "loss": 2.7126,
-      "step": 200500
+      "learning_rate": 1.4217888739635926e-05,
+      "loss": 2.8585,
+      "step": 216300
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4197879000967338e-05,
-      "loss": 2.71,
-      "step": 200600
+      "learning_rate": 1.4201345923519252e-05,
+      "loss": 2.854,
+      "step": 216400
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4180031483021658e-05,
-      "loss": 2.7111,
-      "step": 200700
+      "learning_rate": 1.418480310740258e-05,
+      "loss": 2.8586,
+      "step": 216500
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4162183965075979e-05,
-      "loss": 2.7108,
-      "step": 200800
+      "learning_rate": 1.4168260291285906e-05,
+      "loss": 2.8546,
+      "step": 216600
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.41443364471303e-05,
-      "loss": 2.7228,
-      "step": 200900
+      "learning_rate": 1.4151717475169234e-05,
+      "loss": 2.8522,
+      "step": 216700
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.412648892918462e-05,
-      "loss": 2.7085,
-      "step": 201000
+      "learning_rate": 1.413517465905256e-05,
+      "loss": 2.8486,
+      "step": 216800
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.410864141123894e-05,
-      "loss": 2.7095,
-      "step": 201100
+      "learning_rate": 1.4118631842935887e-05,
+      "loss": 2.8545,
+      "step": 216900
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.409079389329326e-05,
-      "loss": 2.7077,
-      "step": 201200
+      "learning_rate": 1.4102089026819215e-05,
+      "loss": 2.8424,
+      "step": 217000
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4072946375347581e-05,
-      "loss": 2.7039,
-      "step": 201300
+      "learning_rate": 1.408554621070254e-05,
+      "loss": 2.8461,
+      "step": 217100
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4055098857401903e-05,
-      "loss": 2.7041,
-      "step": 201400
+      "learning_rate": 1.4069003394585867e-05,
+      "loss": 2.8562,
+      "step": 217200
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4037251339456222e-05,
-      "loss": 2.7023,
-      "step": 201500
+      "learning_rate": 1.4052460578469195e-05,
+      "loss": 2.8659,
+      "step": 217300
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4019403821510543e-05,
-      "loss": 2.7144,
-      "step": 201600
+      "learning_rate": 1.4035917762352521e-05,
+      "loss": 2.8706,
+      "step": 217400
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.4001556303564865e-05,
-      "loss": 2.7156,
-      "step": 201700
+      "learning_rate": 1.4019374946235847e-05,
+      "loss": 2.8602,
+      "step": 217500
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3983708785619184e-05,
-      "loss": 2.7142,
-      "step": 201800
+      "learning_rate": 1.4002832130119175e-05,
+      "loss": 2.865,
+      "step": 217600
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3965861267673505e-05,
-      "loss": 2.7201,
-      "step": 201900
+      "learning_rate": 1.3986289314002502e-05,
+      "loss": 2.8466,
+      "step": 217700
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3948013749727827e-05,
-      "loss": 2.6971,
-      "step": 202000
+      "learning_rate": 1.396974649788583e-05,
+      "loss": 2.839,
+      "step": 217800
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3930166231782146e-05,
-      "loss": 2.7061,
-      "step": 202100
+      "learning_rate": 1.3953203681769156e-05,
+      "loss": 2.846,
+      "step": 217900
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3912318713836467e-05,
-      "loss": 2.6988,
-      "step": 202200
+      "learning_rate": 1.3936660865652482e-05,
+      "loss": 2.859,
+      "step": 218000
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3894471195890787e-05,
-      "loss": 2.6982,
-      "step": 202300
+      "learning_rate": 1.392011804953581e-05,
+      "loss": 2.8543,
+      "step": 218100
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3876623677945108e-05,
-      "loss": 2.7014,
-      "step": 202400
+      "learning_rate": 1.3903575233419136e-05,
+      "loss": 2.8592,
+      "step": 218200
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.385877615999943e-05,
-      "loss": 2.7116,
-      "step": 202500
+      "learning_rate": 1.3887032417302462e-05,
+      "loss": 2.8545,
+      "step": 218300
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3840928642053749e-05,
-      "loss": 2.7166,
-      "step": 202600
+      "learning_rate": 1.387048960118579e-05,
+      "loss": 2.8506,
+      "step": 218400
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.382308112410807e-05,
-      "loss": 2.7059,
-      "step": 202700
+      "learning_rate": 1.3853946785069116e-05,
+      "loss": 2.8415,
+      "step": 218500
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3805233606162391e-05,
-      "loss": 2.7062,
-      "step": 202800
+      "learning_rate": 1.3837403968952443e-05,
+      "loss": 2.8601,
+      "step": 218600
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.378738608821671e-05,
-      "loss": 2.7098,
-      "step": 202900
+      "learning_rate": 1.382086115283577e-05,
+      "loss": 2.8479,
+      "step": 218700
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3769538570271035e-05,
-      "loss": 2.7037,
-      "step": 203000
+      "learning_rate": 1.3804318336719097e-05,
+      "loss": 2.8401,
+      "step": 218800
     },
     {
       "epoch": 0.72,
-      "learning_rate": 1.3751691052325355e-05,
-      "loss": 2.6999,
-      "step": 203100
+      "learning_rate": 1.3787775520602425e-05,
+      "loss": 2.8496,
+      "step": 218900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3771232704485751e-05,
+      "loss": 2.8539,
+      "step": 219000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3754689888369077e-05,
+      "loss": 2.8554,
+      "step": 219100
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3733843534379676e-05,
-      "loss": 2.7086,
-      "step": 203200
+      "learning_rate": 1.3738147072252405e-05,
+      "loss": 2.8522,
+      "step": 219200
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3715996016433996e-05,
-      "loss": 2.6995,
-      "step": 203300
+      "learning_rate": 1.3721604256135731e-05,
+      "loss": 2.8517,
+      "step": 219300
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3698148498488317e-05,
-      "loss": 2.711,
-      "step": 203400
+      "learning_rate": 1.3705061440019058e-05,
+      "loss": 2.8392,
+      "step": 219400
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3680300980542638e-05,
-      "loss": 2.7133,
-      "step": 203500
+      "learning_rate": 1.3688518623902385e-05,
+      "loss": 2.8428,
+      "step": 219500
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3662453462596957e-05,
-      "loss": 2.7038,
-      "step": 203600
+      "learning_rate": 1.3671975807785712e-05,
+      "loss": 2.8378,
+      "step": 219600
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3644605944651279e-05,
-      "loss": 2.7063,
-      "step": 203700
+      "learning_rate": 1.3655432991669038e-05,
+      "loss": 2.8491,
+      "step": 219700
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.36267584267056e-05,
-      "loss": 2.7157,
-      "step": 203800
+      "learning_rate": 1.3638890175552366e-05,
+      "loss": 2.8435,
+      "step": 219800
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.360891090875992e-05,
-      "loss": 2.7108,
-      "step": 203900
+      "learning_rate": 1.3622347359435692e-05,
+      "loss": 2.8429,
+      "step": 219900
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.359106339081424e-05,
-      "loss": 2.7191,
-      "step": 204000
+      "learning_rate": 1.360580454331902e-05,
+      "loss": 2.857,
+      "step": 220000
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.357321587286856e-05,
-      "loss": 2.7154,
-      "step": 204100
+      "learning_rate": 1.3589261727202346e-05,
+      "loss": 2.8535,
+      "step": 220100
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3555368354922881e-05,
-      "loss": 2.7079,
-      "step": 204200
+      "learning_rate": 1.3572718911085672e-05,
+      "loss": 2.8359,
+      "step": 220200
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3537520836977203e-05,
-      "loss": 2.7155,
-      "step": 204300
+      "learning_rate": 1.3556176094969e-05,
+      "loss": 2.8596,
+      "step": 220300
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3519673319031522e-05,
-      "loss": 2.7104,
-      "step": 204400
+      "learning_rate": 1.3539633278852327e-05,
+      "loss": 2.8304,
+      "step": 220400
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3501825801085843e-05,
-      "loss": 2.7177,
-      "step": 204500
+      "learning_rate": 1.3523090462735651e-05,
+      "loss": 2.861,
+      "step": 220500
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3483978283140165e-05,
-      "loss": 2.7027,
-      "step": 204600
+      "learning_rate": 1.350654764661898e-05,
+      "loss": 2.8567,
+      "step": 220600
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3466130765194484e-05,
-      "loss": 2.7069,
-      "step": 204700
+      "learning_rate": 1.3490004830502307e-05,
+      "loss": 2.8475,
+      "step": 220700
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3448283247248805e-05,
-      "loss": 2.7113,
-      "step": 204800
+      "learning_rate": 1.3473462014385632e-05,
+      "loss": 2.8659,
+      "step": 220800
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3430435729303127e-05,
-      "loss": 2.7089,
-      "step": 204900
+      "learning_rate": 1.3456919198268961e-05,
+      "loss": 2.8468,
+      "step": 220900
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3412588211357446e-05,
-      "loss": 2.7084,
-      "step": 205000
+      "learning_rate": 1.3440376382152287e-05,
+      "loss": 2.8396,
+      "step": 221000
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3394740693411767e-05,
-      "loss": 2.7063,
-      "step": 205100
+      "learning_rate": 1.3423833566035615e-05,
+      "loss": 2.8404,
+      "step": 221100
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3376893175466087e-05,
-      "loss": 2.7047,
-      "step": 205200
+      "learning_rate": 1.3407290749918942e-05,
+      "loss": 2.8522,
+      "step": 221200
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3359045657520408e-05,
-      "loss": 2.7045,
-      "step": 205300
+      "learning_rate": 1.3390747933802266e-05,
+      "loss": 2.8529,
+      "step": 221300
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.334119813957473e-05,
-      "loss": 2.7003,
-      "step": 205400
+      "learning_rate": 1.3374205117685596e-05,
+      "loss": 2.8464,
+      "step": 221400
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3323350621629049e-05,
-      "loss": 2.7127,
-      "step": 205500
+      "learning_rate": 1.3357662301568922e-05,
+      "loss": 2.8546,
+      "step": 221500
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3305503103683373e-05,
-      "loss": 2.706,
-      "step": 205600
+      "learning_rate": 1.3341119485452246e-05,
+      "loss": 2.8414,
+      "step": 221600
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3287655585737693e-05,
-      "loss": 2.7087,
-      "step": 205700
+      "learning_rate": 1.3324576669335576e-05,
+      "loss": 2.8621,
+      "step": 221700
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3269808067792014e-05,
-      "loss": 2.7042,
-      "step": 205800
+      "learning_rate": 1.33080338532189e-05,
+      "loss": 2.8488,
+      "step": 221800
     },
     {
       "epoch": 0.73,
-      "learning_rate": 1.3251960549846335e-05,
-      "loss": 2.7102,
-      "step": 205900
+      "learning_rate": 1.329149103710223e-05,
+      "loss": 2.8321,
+      "step": 221900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3274948220985556e-05,
+      "loss": 2.854,
+      "step": 222000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.3258405404868881e-05,
+      "loss": 2.8554,
+      "step": 222100
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3234113031900655e-05,
-      "loss": 2.7052,
-      "step": 206000
+      "learning_rate": 1.324186258875221e-05,
+      "loss": 2.8565,
+      "step": 222200
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3216265513954976e-05,
-      "loss": 2.7132,
-      "step": 206100
+      "learning_rate": 1.3225319772635537e-05,
+      "loss": 2.8373,
+      "step": 222300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3208776956518861e-05,
+      "loss": 2.8445,
+      "step": 222400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3192234140402191e-05,
+      "loss": 2.8412,
+      "step": 222500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3175691324285516e-05,
+      "loss": 2.8435,
+      "step": 222600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3159148508168842e-05,
+      "loss": 2.8374,
+      "step": 222700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.3142605692052171e-05,
+      "loss": 2.847,
+      "step": 222800
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3198417996009296e-05,
-      "loss": 2.7066,
-      "step": 206200
+      "learning_rate": 1.3126062875935496e-05,
+      "loss": 2.839,
+      "step": 222900
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3180570478063617e-05,
-      "loss": 2.7044,
-      "step": 206300
+      "learning_rate": 1.3109520059818826e-05,
+      "loss": 2.8424,
+      "step": 223000
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3162722960117938e-05,
-      "loss": 2.7034,
-      "step": 206400
+      "learning_rate": 1.309297724370215e-05,
+      "loss": 2.8437,
+      "step": 223100
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3144875442172257e-05,
-      "loss": 2.7098,
-      "step": 206500
+      "learning_rate": 1.3076434427585476e-05,
+      "loss": 2.855,
+      "step": 223200
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3127027924226579e-05,
-      "loss": 2.7157,
-      "step": 206600
+      "learning_rate": 1.3059891611468806e-05,
+      "loss": 2.8544,
+      "step": 223300
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.31091804062809e-05,
-      "loss": 2.7046,
-      "step": 206700
+      "learning_rate": 1.304334879535213e-05,
+      "loss": 2.846,
+      "step": 223400
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.309133288833522e-05,
-      "loss": 2.7127,
-      "step": 206800
+      "learning_rate": 1.3026805979235457e-05,
+      "loss": 2.8462,
+      "step": 223500
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.307348537038954e-05,
-      "loss": 2.7212,
-      "step": 206900
+      "learning_rate": 1.3010263163118786e-05,
+      "loss": 2.8434,
+      "step": 223600
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3055637852443862e-05,
-      "loss": 2.7121,
-      "step": 207000
+      "learning_rate": 1.299372034700211e-05,
+      "loss": 2.8413,
+      "step": 223700
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3037790334498181e-05,
-      "loss": 2.7047,
-      "step": 207100
+      "learning_rate": 1.2977177530885437e-05,
+      "loss": 2.8594,
+      "step": 223800
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3019942816552503e-05,
-      "loss": 2.7158,
-      "step": 207200
+      "learning_rate": 1.2960634714768765e-05,
+      "loss": 2.8467,
+      "step": 223900
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.3002095298606822e-05,
-      "loss": 2.7122,
-      "step": 207300
+      "learning_rate": 1.2944091898652091e-05,
+      "loss": 2.8528,
+      "step": 224000
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2984247780661143e-05,
-      "loss": 2.7048,
-      "step": 207400
+      "learning_rate": 1.292754908253542e-05,
+      "loss": 2.8628,
+      "step": 224100
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2966400262715465e-05,
-      "loss": 2.6977,
-      "step": 207500
+      "learning_rate": 1.2911006266418745e-05,
+      "loss": 2.8545,
+      "step": 224200
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2948552744769784e-05,
-      "loss": 2.7032,
-      "step": 207600
+      "learning_rate": 1.2894463450302072e-05,
+      "loss": 2.8439,
+      "step": 224300
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2930705226824105e-05,
-      "loss": 2.7065,
-      "step": 207700
+      "learning_rate": 1.28779206341854e-05,
+      "loss": 2.8564,
+      "step": 224400
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2912857708878426e-05,
-      "loss": 2.7104,
-      "step": 207800
+      "learning_rate": 1.2861377818068726e-05,
+      "loss": 2.8539,
+      "step": 224500
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2895010190932746e-05,
-      "loss": 2.7086,
-      "step": 207900
+      "learning_rate": 1.2844835001952052e-05,
+      "loss": 2.8458,
+      "step": 224600
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2877162672987067e-05,
-      "loss": 2.7076,
-      "step": 208000
+      "learning_rate": 1.282829218583538e-05,
+      "loss": 2.8507,
+      "step": 224700
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.285931515504139e-05,
-      "loss": 2.7008,
-      "step": 208100
+      "learning_rate": 1.2811749369718706e-05,
+      "loss": 2.8554,
+      "step": 224800
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2841467637095711e-05,
-      "loss": 2.712,
-      "step": 208200
+      "learning_rate": 1.2795206553602032e-05,
+      "loss": 2.858,
+      "step": 224900
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2823620119150031e-05,
-      "loss": 2.7053,
-      "step": 208300
+      "learning_rate": 1.277866373748536e-05,
+      "loss": 2.8638,
+      "step": 225000
     },
     {
       "epoch": 0.74,
-      "learning_rate": 1.2805772601204352e-05,
-      "loss": 2.7051,
-      "step": 208400
+      "learning_rate": 1.2762120921368686e-05,
+      "loss": 2.8561,
+      "step": 225100
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.2787925083258673e-05,
-      "loss": 2.7001,
-      "step": 208500
+      "epoch": 0.75,
+      "learning_rate": 1.2745578105252014e-05,
+      "loss": 2.848,
+      "step": 225200
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.2770077565312993e-05,
-      "loss": 2.697,
-      "step": 208600
+      "epoch": 0.75,
+      "learning_rate": 1.272903528913534e-05,
+      "loss": 2.8534,
+      "step": 225300
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.2752230047367314e-05,
-      "loss": 2.7099,
-      "step": 208700
+      "epoch": 0.75,
+      "learning_rate": 1.2712492473018667e-05,
+      "loss": 2.8433,
+      "step": 225400
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2734382529421635e-05,
-      "loss": 2.7105,
-      "step": 208800
+      "learning_rate": 1.2695949656901995e-05,
+      "loss": 2.8451,
+      "step": 225500
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2716535011475955e-05,
-      "loss": 2.7111,
-      "step": 208900
+      "learning_rate": 1.2679406840785321e-05,
+      "loss": 2.8594,
+      "step": 225600
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2698687493530276e-05,
-      "loss": 2.7154,
-      "step": 209000
+      "learning_rate": 1.2662864024668647e-05,
+      "loss": 2.8397,
+      "step": 225700
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2680839975584596e-05,
-      "loss": 2.7073,
-      "step": 209100
+      "learning_rate": 1.2646321208551975e-05,
+      "loss": 2.8263,
+      "step": 225800
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2662992457638917e-05,
-      "loss": 2.7087,
-      "step": 209200
+      "learning_rate": 1.2629778392435301e-05,
+      "loss": 2.8711,
+      "step": 225900
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2645144939693238e-05,
-      "loss": 2.7062,
-      "step": 209300
+      "learning_rate": 1.2613235576318628e-05,
+      "loss": 2.8541,
+      "step": 226000
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2627297421747557e-05,
-      "loss": 2.7118,
-      "step": 209400
+      "learning_rate": 1.2596692760201956e-05,
+      "loss": 2.8598,
+      "step": 226100
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2609449903801879e-05,
-      "loss": 2.7088,
-      "step": 209500
+      "learning_rate": 1.2580149944085282e-05,
+      "loss": 2.8412,
+      "step": 226200
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.25916023858562e-05,
-      "loss": 2.7063,
-      "step": 209600
+      "learning_rate": 1.256360712796861e-05,
+      "loss": 2.8473,
+      "step": 226300
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.257375486791052e-05,
-      "loss": 2.6977,
-      "step": 209700
+      "learning_rate": 1.2547064311851936e-05,
+      "loss": 2.8354,
+      "step": 226400
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.255590734996484e-05,
-      "loss": 2.704,
-      "step": 209800
+      "learning_rate": 1.2530521495735262e-05,
+      "loss": 2.8492,
+      "step": 226500
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2538059832019162e-05,
-      "loss": 2.7019,
-      "step": 209900
+      "learning_rate": 1.251397867961859e-05,
+      "loss": 2.8495,
+      "step": 226600
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2520212314073481e-05,
-      "loss": 2.7064,
-      "step": 210000
+      "learning_rate": 1.2497435863501916e-05,
+      "loss": 2.8405,
+      "step": 226700
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2502364796127803e-05,
-      "loss": 2.7135,
-      "step": 210100
+      "learning_rate": 1.2480893047385244e-05,
+      "loss": 2.8497,
+      "step": 226800
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2484517278182124e-05,
-      "loss": 2.699,
-      "step": 210200
+      "learning_rate": 1.246435023126857e-05,
+      "loss": 2.841,
+      "step": 226900
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2466669760236445e-05,
-      "loss": 2.6967,
-      "step": 210300
+      "learning_rate": 1.2447807415151897e-05,
+      "loss": 2.8232,
+      "step": 227000
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2448822242290766e-05,
-      "loss": 2.6995,
-      "step": 210400
+      "learning_rate": 1.2431264599035223e-05,
+      "loss": 2.8488,
+      "step": 227100
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2430974724345086e-05,
-      "loss": 2.7137,
-      "step": 210500
+      "learning_rate": 1.241472178291855e-05,
+      "loss": 2.855,
+      "step": 227200
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2413127206399407e-05,
-      "loss": 2.7117,
-      "step": 210600
+      "learning_rate": 1.2398178966801877e-05,
+      "loss": 2.8512,
+      "step": 227300
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2395279688453726e-05,
-      "loss": 2.7145,
-      "step": 210700
+      "learning_rate": 1.2381636150685203e-05,
+      "loss": 2.851,
+      "step": 227400
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2377432170508048e-05,
-      "loss": 2.7083,
-      "step": 210800
+      "learning_rate": 1.2365093334568531e-05,
+      "loss": 2.8509,
+      "step": 227500
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2359584652562369e-05,
-      "loss": 2.7017,
-      "step": 210900
+      "learning_rate": 1.2348550518451857e-05,
+      "loss": 2.8454,
+      "step": 227600
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2341737134616688e-05,
-      "loss": 2.7052,
-      "step": 211000
+      "learning_rate": 1.2332007702335184e-05,
+      "loss": 2.8467,
+      "step": 227700
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.232388961667101e-05,
-      "loss": 2.7031,
-      "step": 211100
+      "learning_rate": 1.2315464886218512e-05,
+      "loss": 2.8355,
+      "step": 227800
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2306042098725331e-05,
-      "loss": 2.6986,
-      "step": 211200
+      "learning_rate": 1.2298922070101838e-05,
+      "loss": 2.8411,
+      "step": 227900
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2288194580779652e-05,
-      "loss": 2.7134,
-      "step": 211300
+      "learning_rate": 1.2282379253985166e-05,
+      "loss": 2.8533,
+      "step": 228000
     },
     {
       "epoch": 0.75,
-      "learning_rate": 1.2270347062833973e-05,
-      "loss": 2.7121,
-      "step": 211400
+      "learning_rate": 1.2265836437868492e-05,
+      "loss": 2.8462,
+      "step": 228100
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.2252499544888293e-05,
-      "loss": 2.6999,
-      "step": 211500
+      "epoch": 0.76,
+      "learning_rate": 1.2249293621751818e-05,
+      "loss": 2.8534,
+      "step": 228200
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2234652026942614e-05,
-      "loss": 2.7063,
-      "step": 211600
+      "learning_rate": 1.2232750805635146e-05,
+      "loss": 2.8544,
+      "step": 228300
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2216804508996935e-05,
-      "loss": 2.7068,
-      "step": 211700
+      "learning_rate": 1.2216207989518472e-05,
+      "loss": 2.8454,
+      "step": 228400
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2198956991051255e-05,
-      "loss": 2.711,
-      "step": 211800
+      "learning_rate": 1.2199665173401799e-05,
+      "loss": 2.8486,
+      "step": 228500
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2181109473105576e-05,
-      "loss": 2.7179,
-      "step": 211900
+      "learning_rate": 1.2183122357285127e-05,
+      "loss": 2.8304,
+      "step": 228600
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2163261955159897e-05,
-      "loss": 2.7077,
-      "step": 212000
+      "learning_rate": 1.2166579541168453e-05,
+      "loss": 2.85,
+      "step": 228700
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2145414437214217e-05,
-      "loss": 2.703,
-      "step": 212100
+      "learning_rate": 1.2150036725051779e-05,
+      "loss": 2.8406,
+      "step": 228800
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2127566919268538e-05,
-      "loss": 2.709,
-      "step": 212200
+      "learning_rate": 1.2133493908935107e-05,
+      "loss": 2.8612,
+      "step": 228900
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2109719401322857e-05,
-      "loss": 2.703,
-      "step": 212300
+      "learning_rate": 1.2116951092818433e-05,
+      "loss": 2.8448,
+      "step": 229000
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2091871883377179e-05,
-      "loss": 2.7035,
-      "step": 212400
+      "learning_rate": 1.2100408276701761e-05,
+      "loss": 2.8338,
+      "step": 229100
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.20740243654315e-05,
-      "loss": 2.6925,
-      "step": 212500
+      "learning_rate": 1.2083865460585087e-05,
+      "loss": 2.8535,
+      "step": 229200
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2056176847485821e-05,
-      "loss": 2.7145,
-      "step": 212600
+      "learning_rate": 1.2067322644468413e-05,
+      "loss": 2.8487,
+      "step": 229300
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2038329329540142e-05,
-      "loss": 2.7073,
-      "step": 212700
+      "learning_rate": 1.2050779828351741e-05,
+      "loss": 2.8289,
+      "step": 229400
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2020481811594462e-05,
-      "loss": 2.7192,
-      "step": 212800
+      "learning_rate": 1.2034237012235066e-05,
+      "loss": 2.8515,
+      "step": 229500
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.2002634293648783e-05,
-      "loss": 2.7103,
-      "step": 212900
+      "learning_rate": 1.2017694196118394e-05,
+      "loss": 2.8513,
+      "step": 229600
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1984786775703104e-05,
-      "loss": 2.7006,
-      "step": 213000
+      "learning_rate": 1.2001151380001722e-05,
+      "loss": 2.8516,
+      "step": 229700
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1966939257757424e-05,
-      "loss": 2.7086,
-      "step": 213100
+      "learning_rate": 1.1984608563885048e-05,
+      "loss": 2.8506,
+      "step": 229800
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1949091739811745e-05,
-      "loss": 2.7046,
-      "step": 213200
+      "learning_rate": 1.1968065747768374e-05,
+      "loss": 2.8482,
+      "step": 229900
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1931244221866066e-05,
-      "loss": 2.7126,
-      "step": 213300
+      "learning_rate": 1.1951522931651702e-05,
+      "loss": 2.8488,
+      "step": 230000
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1913396703920386e-05,
-      "loss": 2.6996,
-      "step": 213400
+      "learning_rate": 1.1934980115535028e-05,
+      "loss": 2.8363,
+      "step": 230100
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1895549185974707e-05,
-      "loss": 2.7086,
-      "step": 213500
+      "learning_rate": 1.1918437299418356e-05,
+      "loss": 2.8563,
+      "step": 230200
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1877701668029028e-05,
-      "loss": 2.7119,
-      "step": 213600
+      "learning_rate": 1.1901894483301681e-05,
+      "loss": 2.8554,
+      "step": 230300
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1859854150083348e-05,
-      "loss": 2.6988,
-      "step": 213700
+      "learning_rate": 1.1885351667185009e-05,
+      "loss": 2.8617,
+      "step": 230400
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1842006632137669e-05,
-      "loss": 2.7048,
-      "step": 213800
+      "learning_rate": 1.1868808851068337e-05,
+      "loss": 2.8383,
+      "step": 230500
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.182415911419199e-05,
-      "loss": 2.7112,
-      "step": 213900
+      "learning_rate": 1.1852266034951661e-05,
+      "loss": 2.8433,
+      "step": 230600
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1806311596246311e-05,
-      "loss": 2.7044,
-      "step": 214000
+      "learning_rate": 1.183572321883499e-05,
+      "loss": 2.8385,
+      "step": 230700
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1788464078300633e-05,
-      "loss": 2.7009,
-      "step": 214100
+      "learning_rate": 1.1819180402718315e-05,
+      "loss": 2.8447,
+      "step": 230800
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1770616560354952e-05,
-      "loss": 2.7011,
-      "step": 214200
+      "learning_rate": 1.1802637586601643e-05,
+      "loss": 2.8499,
+      "step": 230900
     },
     {
       "epoch": 0.76,
-      "learning_rate": 1.1752769042409273e-05,
-      "loss": 2.7038,
-      "step": 214300
+      "learning_rate": 1.178609477048497e-05,
+      "loss": 2.8552,
+      "step": 231000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1769551954368296e-05,
+      "loss": 2.8315,
+      "step": 231100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.1753009138251624e-05,
+      "loss": 2.8357,
+      "step": 231200
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1734921524463593e-05,
-      "loss": 2.6994,
-      "step": 214400
+      "learning_rate": 1.1736466322134952e-05,
+      "loss": 2.8564,
+      "step": 231300
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1717074006517914e-05,
-      "loss": 2.7019,
-      "step": 214500
+      "learning_rate": 1.1719923506018276e-05,
+      "loss": 2.8525,
+      "step": 231400
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1699226488572235e-05,
-      "loss": 2.7028,
-      "step": 214600
+      "learning_rate": 1.1703380689901604e-05,
+      "loss": 2.8418,
+      "step": 231500
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1681378970626555e-05,
-      "loss": 2.7122,
-      "step": 214700
+      "learning_rate": 1.168683787378493e-05,
+      "loss": 2.8755,
+      "step": 231600
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1663531452680876e-05,
-      "loss": 2.7032,
-      "step": 214800
+      "learning_rate": 1.1670295057668258e-05,
+      "loss": 2.8496,
+      "step": 231700
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1645683934735197e-05,
-      "loss": 2.6944,
-      "step": 214900
+      "learning_rate": 1.1653752241551584e-05,
+      "loss": 2.8552,
+      "step": 231800
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1627836416789517e-05,
-      "loss": 2.7105,
-      "step": 215000
+      "learning_rate": 1.163720942543491e-05,
+      "loss": 2.8271,
+      "step": 231900
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1609988898843838e-05,
-      "loss": 2.6958,
-      "step": 215100
+      "learning_rate": 1.1620666609318239e-05,
+      "loss": 2.8499,
+      "step": 232000
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1592141380898159e-05,
-      "loss": 2.701,
-      "step": 215200
+      "learning_rate": 1.1604123793201565e-05,
+      "loss": 2.848,
+      "step": 232100
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.157429386295248e-05,
-      "loss": 2.7047,
-      "step": 215300
+      "learning_rate": 1.1587580977084891e-05,
+      "loss": 2.8471,
+      "step": 232200
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1556446345006802e-05,
-      "loss": 2.7097,
-      "step": 215400
+      "learning_rate": 1.1571038160968219e-05,
+      "loss": 2.8432,
+      "step": 232300
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1538598827061121e-05,
-      "loss": 2.6957,
-      "step": 215500
+      "learning_rate": 1.1554495344851545e-05,
+      "loss": 2.8658,
+      "step": 232400
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1520751309115442e-05,
-      "loss": 2.7055,
-      "step": 215600
+      "learning_rate": 1.1537952528734871e-05,
+      "loss": 2.8563,
+      "step": 232500
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1502903791169762e-05,
-      "loss": 2.7012,
-      "step": 215700
+      "learning_rate": 1.15214097126182e-05,
+      "loss": 2.847,
+      "step": 232600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1504866896501526e-05,
+      "loss": 2.8411,
+      "step": 232700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1488324080384854e-05,
+      "loss": 2.8445,
+      "step": 232800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.147178126426818e-05,
+      "loss": 2.8421,
+      "step": 232900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.1455238448151506e-05,
+      "loss": 2.8559,
+      "step": 233000
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1485056273224083e-05,
-      "loss": 2.7047,
-      "step": 215800
+      "learning_rate": 1.1438695632034834e-05,
+      "loss": 2.8482,
+      "step": 233100
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1467208755278404e-05,
-      "loss": 2.6991,
-      "step": 215900
+      "learning_rate": 1.1422152815918158e-05,
+      "loss": 2.8498,
+      "step": 233200
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1449361237332724e-05,
-      "loss": 2.7148,
-      "step": 216000
+      "learning_rate": 1.1405609999801486e-05,
+      "loss": 2.8407,
+      "step": 233300
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1431513719387045e-05,
-      "loss": 2.6978,
-      "step": 216100
+      "learning_rate": 1.1389067183684814e-05,
+      "loss": 2.8554,
+      "step": 233400
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1413666201441366e-05,
-      "loss": 2.6998,
-      "step": 216200
+      "learning_rate": 1.137252436756814e-05,
+      "loss": 2.8531,
+      "step": 233500
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1395818683495686e-05,
-      "loss": 2.7136,
-      "step": 216300
+      "learning_rate": 1.1355981551451467e-05,
+      "loss": 2.8443,
+      "step": 233600
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1377971165550009e-05,
-      "loss": 2.6882,
-      "step": 216400
+      "learning_rate": 1.1339438735334795e-05,
+      "loss": 2.852,
+      "step": 233700
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1360123647604328e-05,
-      "loss": 2.7058,
-      "step": 216500
+      "learning_rate": 1.1322895919218121e-05,
+      "loss": 2.8438,
+      "step": 233800
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.134227612965865e-05,
-      "loss": 2.7071,
-      "step": 216600
+      "learning_rate": 1.1306353103101449e-05,
+      "loss": 2.8466,
+      "step": 233900
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.132442861171297e-05,
-      "loss": 2.6967,
-      "step": 216700
+      "learning_rate": 1.1289810286984773e-05,
+      "loss": 2.845,
+      "step": 234000
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.130658109376729e-05,
-      "loss": 2.6981,
-      "step": 216800
+      "learning_rate": 1.1273267470868101e-05,
+      "loss": 2.8367,
+      "step": 234100
     },
     {
       "epoch": 0.77,
-      "learning_rate": 1.1288733575821611e-05,
-      "loss": 2.6987,
-      "step": 216900
+      "learning_rate": 1.125672465475143e-05,
+      "loss": 2.853,
+      "step": 234200
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.1270886057875933e-05,
-      "loss": 2.7027,
-      "step": 217000
+      "epoch": 0.78,
+      "learning_rate": 1.1240181838634754e-05,
+      "loss": 2.8384,
+      "step": 234300
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.1253038539930252e-05,
-      "loss": 2.7022,
-      "step": 217100
+      "epoch": 0.78,
+      "learning_rate": 1.1223639022518082e-05,
+      "loss": 2.8399,
+      "step": 234400
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1235191021984573e-05,
-      "loss": 2.7012,
-      "step": 217200
+      "learning_rate": 1.1207096206401408e-05,
+      "loss": 2.8486,
+      "step": 234500
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1217343504038893e-05,
-      "loss": 2.6959,
-      "step": 217300
+      "learning_rate": 1.1190553390284736e-05,
+      "loss": 2.8401,
+      "step": 234600
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1199495986093214e-05,
-      "loss": 2.7083,
-      "step": 217400
+      "learning_rate": 1.1174010574168062e-05,
+      "loss": 2.843,
+      "step": 234700
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1181648468147535e-05,
-      "loss": 2.705,
-      "step": 217500
+      "learning_rate": 1.1157467758051388e-05,
+      "loss": 2.8443,
+      "step": 234800
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1163800950201855e-05,
-      "loss": 2.7049,
-      "step": 217600
+      "learning_rate": 1.1140924941934716e-05,
+      "loss": 2.8469,
+      "step": 234900
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1145953432256178e-05,
-      "loss": 2.7069,
-      "step": 217700
+      "learning_rate": 1.1124382125818044e-05,
+      "loss": 2.8405,
+      "step": 235000
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1128105914310497e-05,
-      "loss": 2.7007,
-      "step": 217800
+      "learning_rate": 1.1107839309701369e-05,
+      "loss": 2.8459,
+      "step": 235100
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1110258396364818e-05,
-      "loss": 2.7038,
-      "step": 217900
+      "learning_rate": 1.1091296493584697e-05,
+      "loss": 2.8383,
+      "step": 235200
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.109241087841914e-05,
-      "loss": 2.7036,
-      "step": 218000
+      "learning_rate": 1.1074753677468023e-05,
+      "loss": 2.8333,
+      "step": 235300
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1074563360473459e-05,
-      "loss": 2.7045,
-      "step": 218100
+      "learning_rate": 1.105821086135135e-05,
+      "loss": 2.8358,
+      "step": 235400
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.105671584252778e-05,
-      "loss": 2.7068,
-      "step": 218200
+      "learning_rate": 1.1041668045234677e-05,
+      "loss": 2.8275,
+      "step": 235500
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1038868324582102e-05,
-      "loss": 2.6995,
-      "step": 218300
+      "learning_rate": 1.1025125229118003e-05,
+      "loss": 2.8276,
+      "step": 235600
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1021020806636421e-05,
-      "loss": 2.7003,
-      "step": 218400
+      "learning_rate": 1.1008582413001331e-05,
+      "loss": 2.8447,
+      "step": 235700
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.1003173288690742e-05,
-      "loss": 2.7071,
-      "step": 218500
+      "learning_rate": 1.0992039596884657e-05,
+      "loss": 2.8347,
+      "step": 235800
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0985325770745063e-05,
-      "loss": 2.7033,
-      "step": 218600
+      "learning_rate": 1.0975496780767984e-05,
+      "loss": 2.8537,
+      "step": 235900
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0967478252799383e-05,
-      "loss": 2.7064,
-      "step": 218700
+      "learning_rate": 1.0958953964651311e-05,
+      "loss": 2.8527,
+      "step": 236000
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0949630734853704e-05,
-      "loss": 2.6959,
-      "step": 218800
+      "learning_rate": 1.0942411148534638e-05,
+      "loss": 2.8456,
+      "step": 236100
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0931783216908024e-05,
-      "loss": 2.6931,
-      "step": 218900
+      "learning_rate": 1.0925868332417964e-05,
+      "loss": 2.8555,
+      "step": 236200
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0913935698962347e-05,
-      "loss": 2.7059,
-      "step": 219000
+      "learning_rate": 1.0909325516301292e-05,
+      "loss": 2.8431,
+      "step": 236300
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0896088181016668e-05,
-      "loss": 2.6962,
-      "step": 219100
+      "learning_rate": 1.0892782700184618e-05,
+      "loss": 2.8367,
+      "step": 236400
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0878240663070987e-05,
-      "loss": 2.7064,
-      "step": 219200
+      "learning_rate": 1.0876239884067946e-05,
+      "loss": 2.8366,
+      "step": 236500
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0860393145125309e-05,
-      "loss": 2.7002,
-      "step": 219300
+      "learning_rate": 1.0859697067951272e-05,
+      "loss": 2.838,
+      "step": 236600
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0842545627179628e-05,
-      "loss": 2.7041,
-      "step": 219400
+      "learning_rate": 1.0843154251834598e-05,
+      "loss": 2.8432,
+      "step": 236700
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.082469810923395e-05,
-      "loss": 2.7046,
-      "step": 219500
+      "learning_rate": 1.0826611435717926e-05,
+      "loss": 2.8475,
+      "step": 236800
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.080685059128827e-05,
-      "loss": 2.6958,
-      "step": 219600
+      "learning_rate": 1.0810068619601253e-05,
+      "loss": 2.8599,
+      "step": 236900
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.078900307334259e-05,
-      "loss": 2.7129,
-      "step": 219700
+      "learning_rate": 1.0793525803484579e-05,
+      "loss": 2.8542,
+      "step": 237000
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0771155555396911e-05,
-      "loss": 2.7022,
-      "step": 219800
+      "learning_rate": 1.0776982987367907e-05,
+      "loss": 2.856,
+      "step": 237100
     },
     {
       "epoch": 0.78,
-      "learning_rate": 1.0753308037451233e-05,
-      "loss": 2.7013,
-      "step": 219900
+      "learning_rate": 1.0760440171251233e-05,
+      "loss": 2.8454,
+      "step": 237200
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0735460519505552e-05,
-      "loss": 2.6996,
-      "step": 220000
+      "learning_rate": 1.074389735513456e-05,
+      "loss": 2.8457,
+      "step": 237300
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0717613001559873e-05,
-      "loss": 2.7058,
-      "step": 220100
+      "learning_rate": 1.0727354539017887e-05,
+      "loss": 2.848,
+      "step": 237400
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0699765483614194e-05,
-      "loss": 2.7006,
-      "step": 220200
+      "learning_rate": 1.0710811722901213e-05,
+      "loss": 2.8613,
+      "step": 237500
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0681917965668516e-05,
-      "loss": 2.7022,
-      "step": 220300
+      "learning_rate": 1.0694268906784541e-05,
+      "loss": 2.8439,
+      "step": 237600
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0664070447722837e-05,
-      "loss": 2.7062,
-      "step": 220400
+      "learning_rate": 1.0677726090667866e-05,
+      "loss": 2.8428,
+      "step": 237700
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0646222929777156e-05,
-      "loss": 2.6988,
-      "step": 220500
+      "learning_rate": 1.0661183274551194e-05,
+      "loss": 2.8429,
+      "step": 237800
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0628375411831478e-05,
-      "loss": 2.7036,
-      "step": 220600
+      "learning_rate": 1.0644640458434522e-05,
+      "loss": 2.8543,
+      "step": 237900
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0610527893885797e-05,
-      "loss": 2.6986,
-      "step": 220700
+      "learning_rate": 1.0628097642317846e-05,
+      "loss": 2.8352,
+      "step": 238000
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0592680375940118e-05,
-      "loss": 2.7137,
-      "step": 220800
+      "learning_rate": 1.0611554826201174e-05,
+      "loss": 2.8466,
+      "step": 238100
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.057483285799444e-05,
-      "loss": 2.7025,
-      "step": 220900
+      "learning_rate": 1.0595012010084502e-05,
+      "loss": 2.8493,
+      "step": 238200
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0556985340048759e-05,
-      "loss": 2.7126,
-      "step": 221000
+      "learning_rate": 1.0578469193967828e-05,
+      "loss": 2.8519,
+      "step": 238300
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.053913782210308e-05,
-      "loss": 2.7058,
-      "step": 221100
+      "learning_rate": 1.0561926377851155e-05,
+      "loss": 2.8394,
+      "step": 238400
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0521290304157402e-05,
-      "loss": 2.7087,
-      "step": 221200
+      "learning_rate": 1.054538356173448e-05,
+      "loss": 2.8483,
+      "step": 238500
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0503442786211721e-05,
-      "loss": 2.6998,
-      "step": 221300
+      "learning_rate": 1.0528840745617809e-05,
+      "loss": 2.8429,
+      "step": 238600
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0485595268266042e-05,
-      "loss": 2.7031,
-      "step": 221400
+      "learning_rate": 1.0512297929501137e-05,
+      "loss": 2.8508,
+      "step": 238700
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0467747750320363e-05,
-      "loss": 2.7069,
-      "step": 221500
+      "learning_rate": 1.0495755113384461e-05,
+      "loss": 2.8358,
+      "step": 238800
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0449900232374685e-05,
-      "loss": 2.708,
-      "step": 221600
+      "learning_rate": 1.0479212297267789e-05,
+      "loss": 2.8305,
+      "step": 238900
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0432052714429006e-05,
-      "loss": 2.7032,
-      "step": 221700
+      "learning_rate": 1.0462669481151115e-05,
+      "loss": 2.8415,
+      "step": 239000
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0414205196483325e-05,
-      "loss": 2.7039,
-      "step": 221800
+      "learning_rate": 1.0446126665034443e-05,
+      "loss": 2.8509,
+      "step": 239100
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0396357678537647e-05,
-      "loss": 2.6945,
-      "step": 221900
+      "learning_rate": 1.042958384891777e-05,
+      "loss": 2.8471,
+      "step": 239200
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0378510160591968e-05,
-      "loss": 2.6955,
-      "step": 222000
+      "learning_rate": 1.0413041032801096e-05,
+      "loss": 2.8454,
+      "step": 239300
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0360662642646287e-05,
-      "loss": 2.6966,
-      "step": 222100
+      "learning_rate": 1.0396498216684424e-05,
+      "loss": 2.8649,
+      "step": 239400
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0342815124700609e-05,
-      "loss": 2.6987,
-      "step": 222200
+      "learning_rate": 1.037995540056775e-05,
+      "loss": 2.8397,
+      "step": 239500
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0324967606754928e-05,
-      "loss": 2.6986,
-      "step": 222300
+      "learning_rate": 1.0363412584451076e-05,
+      "loss": 2.8395,
+      "step": 239600
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.030712008880925e-05,
-      "loss": 2.7068,
-      "step": 222400
+      "learning_rate": 1.0346869768334404e-05,
+      "loss": 2.8352,
+      "step": 239700
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.028927257086357e-05,
-      "loss": 2.7002,
-      "step": 222500
+      "learning_rate": 1.033032695221773e-05,
+      "loss": 2.8313,
+      "step": 239800
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.027142505291789e-05,
-      "loss": 2.6913,
-      "step": 222600
+      "learning_rate": 1.0313784136101056e-05,
+      "loss": 2.848,
+      "step": 239900
     },
     {
       "epoch": 0.79,
-      "learning_rate": 1.0253577534972211e-05,
-      "loss": 2.6961,
-      "step": 222700
+      "learning_rate": 1.0297241319984384e-05,
+      "loss": 2.8409,
+      "step": 240000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.028069850386771e-05,
+      "loss": 2.8426,
+      "step": 240100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0264155687751038e-05,
+      "loss": 2.8338,
+      "step": 240200
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0235730017026532e-05,
-      "loss": 2.6971,
-      "step": 222800
+      "learning_rate": 1.0247612871634365e-05,
+      "loss": 2.8343,
+      "step": 240300
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0217882499080854e-05,
-      "loss": 2.7025,
-      "step": 222900
+      "learning_rate": 1.0231070055517691e-05,
+      "loss": 2.8419,
+      "step": 240400
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0200034981135175e-05,
-      "loss": 2.7009,
-      "step": 223000
+      "learning_rate": 1.0214527239401019e-05,
+      "loss": 2.8443,
+      "step": 240500
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0182187463189494e-05,
-      "loss": 2.698,
-      "step": 223100
+      "learning_rate": 1.0197984423284345e-05,
+      "loss": 2.856,
+      "step": 240600
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0164339945243816e-05,
-      "loss": 2.6932,
-      "step": 223200
+      "learning_rate": 1.0181441607167671e-05,
+      "loss": 2.862,
+      "step": 240700
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0146492427298137e-05,
-      "loss": 2.6892,
-      "step": 223300
+      "learning_rate": 1.0164898791051e-05,
+      "loss": 2.8492,
+      "step": 240800
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0128644909352456e-05,
-      "loss": 2.7013,
-      "step": 223400
+      "learning_rate": 1.0148355974934325e-05,
+      "loss": 2.8413,
+      "step": 240900
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0110797391406778e-05,
-      "loss": 2.6979,
-      "step": 223500
+      "learning_rate": 1.0131813158817652e-05,
+      "loss": 2.8502,
+      "step": 241000
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0092949873461099e-05,
-      "loss": 2.6968,
-      "step": 223600
+      "learning_rate": 1.011527034270098e-05,
+      "loss": 2.8387,
+      "step": 241100
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0075102355515418e-05,
-      "loss": 2.7025,
-      "step": 223700
+      "learning_rate": 1.0098727526584306e-05,
+      "loss": 2.8535,
+      "step": 241200
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.005725483756974e-05,
-      "loss": 2.7071,
-      "step": 223800
+      "learning_rate": 1.0082184710467634e-05,
+      "loss": 2.8372,
+      "step": 241300
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0039407319624059e-05,
-      "loss": 2.7038,
-      "step": 223900
+      "learning_rate": 1.006564189435096e-05,
+      "loss": 2.8247,
+      "step": 241400
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.002155980167838e-05,
-      "loss": 2.6933,
-      "step": 224000
+      "learning_rate": 1.0049099078234286e-05,
+      "loss": 2.8421,
+      "step": 241500
     },
     {
       "epoch": 0.8,
-      "learning_rate": 1.0003712283732703e-05,
-      "loss": 2.6946,
-      "step": 224100
+      "learning_rate": 1.0032556262117614e-05,
+      "loss": 2.8323,
+      "step": 241600
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.985864765787023e-06,
-      "loss": 2.7156,
-      "step": 224200
+      "learning_rate": 1.0016013446000939e-05,
+      "loss": 2.8521,
+      "step": 241700
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.968017247841344e-06,
-      "loss": 2.7014,
-      "step": 224300
+      "learning_rate": 9.999470629884267e-06,
+      "loss": 2.8413,
+      "step": 241800
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.950169729895663e-06,
-      "loss": 2.7004,
-      "step": 224400
+      "learning_rate": 9.982927813767595e-06,
+      "loss": 2.8555,
+      "step": 241900
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.932322211949985e-06,
-      "loss": 2.7012,
-      "step": 224500
+      "learning_rate": 9.96638499765092e-06,
+      "loss": 2.8447,
+      "step": 242000
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.914474694004306e-06,
-      "loss": 2.7054,
-      "step": 224600
+      "learning_rate": 9.949842181534247e-06,
+      "loss": 2.8486,
+      "step": 242100
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.896627176058625e-06,
-      "loss": 2.6914,
-      "step": 224700
+      "learning_rate": 9.933299365417573e-06,
+      "loss": 2.856,
+      "step": 242200
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.878779658112947e-06,
-      "loss": 2.7055,
-      "step": 224800
+      "learning_rate": 9.916756549300901e-06,
+      "loss": 2.8335,
+      "step": 242300
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.860932140167268e-06,
-      "loss": 2.6997,
-      "step": 224900
+      "learning_rate": 9.900213733184229e-06,
+      "loss": 2.8277,
+      "step": 242400
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.843084622221587e-06,
-      "loss": 2.7085,
-      "step": 225000
+      "learning_rate": 9.883670917067554e-06,
+      "loss": 2.8461,
+      "step": 242500
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.825237104275909e-06,
-      "loss": 2.6912,
-      "step": 225100
+      "learning_rate": 9.867128100950882e-06,
+      "loss": 2.8505,
+      "step": 242600
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.80738958633023e-06,
-      "loss": 2.7002,
-      "step": 225200
+      "learning_rate": 9.85058528483421e-06,
+      "loss": 2.8248,
+      "step": 242700
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.78954206838455e-06,
-      "loss": 2.6993,
-      "step": 225300
+      "learning_rate": 9.834042468717536e-06,
+      "loss": 2.8475,
+      "step": 242800
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.771694550438872e-06,
-      "loss": 2.6978,
-      "step": 225400
+      "learning_rate": 9.817499652600862e-06,
+      "loss": 2.8381,
+      "step": 242900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.800956836484188e-06,
+      "loss": 2.8466,
+      "step": 243000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.784414020367516e-06,
+      "loss": 2.8365,
+      "step": 243100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.767871204250842e-06,
+      "loss": 2.8607,
+      "step": 243200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.751328388134169e-06,
+      "loss": 2.8417,
+      "step": 243300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.734785572017496e-06,
+      "loss": 2.8378,
+      "step": 243400
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 9.753847032493192e-06,
-      "loss": 2.7073,
-      "step": 225500
+      "epoch": 0.81,
+      "learning_rate": 9.718242755900823e-06,
+      "loss": 2.8485,
+      "step": 243500
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.735999514547513e-06,
-      "loss": 2.6984,
-      "step": 225600
+      "learning_rate": 9.701699939784149e-06,
+      "loss": 2.8345,
+      "step": 243600
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.718151996601834e-06,
-      "loss": 2.6929,
-      "step": 225700
+      "learning_rate": 9.685157123667477e-06,
+      "loss": 2.8469,
+      "step": 243700
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.700304478656154e-06,
-      "loss": 2.6988,
-      "step": 225800
+      "learning_rate": 9.668614307550803e-06,
+      "loss": 2.8284,
+      "step": 243800
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.682456960710475e-06,
-      "loss": 2.6957,
-      "step": 225900
+      "learning_rate": 9.652071491434131e-06,
+      "loss": 2.8452,
+      "step": 243900
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.664609442764794e-06,
-      "loss": 2.7019,
-      "step": 226000
+      "learning_rate": 9.635528675317457e-06,
+      "loss": 2.8284,
+      "step": 244000
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.646761924819116e-06,
-      "loss": 2.7073,
-      "step": 226100
+      "learning_rate": 9.618985859200783e-06,
+      "loss": 2.8403,
+      "step": 244100
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.628914406873437e-06,
-      "loss": 2.7022,
-      "step": 226200
+      "learning_rate": 9.602443043084111e-06,
+      "loss": 2.8471,
+      "step": 244200
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.611066888927756e-06,
-      "loss": 2.7146,
-      "step": 226300
+      "learning_rate": 9.585900226967438e-06,
+      "loss": 2.8522,
+      "step": 244300
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.593219370982078e-06,
-      "loss": 2.6932,
-      "step": 226400
+      "learning_rate": 9.569357410850764e-06,
+      "loss": 2.8324,
+      "step": 244400
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.575371853036399e-06,
-      "loss": 2.6939,
-      "step": 226500
+      "learning_rate": 9.552814594734092e-06,
+      "loss": 2.8303,
+      "step": 244500
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.557524335090718e-06,
-      "loss": 2.6955,
-      "step": 226600
+      "learning_rate": 9.536271778617418e-06,
+      "loss": 2.8376,
+      "step": 244600
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.539676817145041e-06,
-      "loss": 2.6939,
-      "step": 226700
+      "learning_rate": 9.519728962500744e-06,
+      "loss": 2.8394,
+      "step": 244700
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.52182929919936e-06,
-      "loss": 2.6939,
-      "step": 226800
+      "learning_rate": 9.503186146384072e-06,
+      "loss": 2.8362,
+      "step": 244800
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.503981781253682e-06,
-      "loss": 2.7074,
-      "step": 226900
+      "learning_rate": 9.486643330267398e-06,
+      "loss": 2.8398,
+      "step": 244900
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.486134263308003e-06,
-      "loss": 2.7026,
-      "step": 227000
+      "learning_rate": 9.470100514150726e-06,
+      "loss": 2.8295,
+      "step": 245000
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.468286745362323e-06,
-      "loss": 2.7055,
-      "step": 227100
+      "learning_rate": 9.453557698034052e-06,
+      "loss": 2.8315,
+      "step": 245100
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.450439227416644e-06,
-      "loss": 2.6944,
-      "step": 227200
+      "learning_rate": 9.437014881917379e-06,
+      "loss": 2.8428,
+      "step": 245200
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.432591709470963e-06,
-      "loss": 2.6982,
-      "step": 227300
+      "learning_rate": 9.420472065800707e-06,
+      "loss": 2.8394,
+      "step": 245300
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.414744191525285e-06,
-      "loss": 2.6879,
-      "step": 227400
+      "learning_rate": 9.403929249684031e-06,
+      "loss": 2.8427,
+      "step": 245400
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.396896673579606e-06,
-      "loss": 2.7001,
-      "step": 227500
+      "learning_rate": 9.387386433567359e-06,
+      "loss": 2.852,
+      "step": 245500
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.379049155633925e-06,
-      "loss": 2.704,
-      "step": 227600
+      "learning_rate": 9.370843617450687e-06,
+      "loss": 2.8627,
+      "step": 245600
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.361201637688247e-06,
-      "loss": 2.6901,
-      "step": 227700
+      "learning_rate": 9.354300801334013e-06,
+      "loss": 2.8475,
+      "step": 245700
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.343354119742568e-06,
-      "loss": 2.7065,
-      "step": 227800
+      "learning_rate": 9.33775798521734e-06,
+      "loss": 2.8473,
+      "step": 245800
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.325506601796889e-06,
-      "loss": 2.696,
-      "step": 227900
+      "learning_rate": 9.321215169100667e-06,
+      "loss": 2.8502,
+      "step": 245900
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.30765908385121e-06,
-      "loss": 2.7024,
-      "step": 228000
+      "learning_rate": 9.304672352983994e-06,
+      "loss": 2.8339,
+      "step": 246000
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.28981156590553e-06,
-      "loss": 2.7057,
-      "step": 228100
+      "learning_rate": 9.288129536867322e-06,
+      "loss": 2.8507,
+      "step": 246100
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.271964047959851e-06,
-      "loss": 2.6975,
-      "step": 228200
+      "learning_rate": 9.271586720750646e-06,
+      "loss": 2.834,
+      "step": 246200
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.254116530014172e-06,
-      "loss": 2.7037,
-      "step": 228300
+      "learning_rate": 9.255043904633974e-06,
+      "loss": 2.843,
+      "step": 246300
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.236269012068492e-06,
-      "loss": 2.6997,
-      "step": 228400
+      "learning_rate": 9.238501088517302e-06,
+      "loss": 2.8389,
+      "step": 246400
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.218421494122813e-06,
-      "loss": 2.6944,
-      "step": 228500
+      "learning_rate": 9.221958272400628e-06,
+      "loss": 2.8441,
+      "step": 246500
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.200573976177134e-06,
-      "loss": 2.6946,
-      "step": 228600
+      "learning_rate": 9.205415456283954e-06,
+      "loss": 2.8534,
+      "step": 246600
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.182726458231454e-06,
-      "loss": 2.7034,
-      "step": 228700
+      "learning_rate": 9.18887264016728e-06,
+      "loss": 2.8622,
+      "step": 246700
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.164878940285775e-06,
-      "loss": 2.7105,
-      "step": 228800
+      "learning_rate": 9.172329824050609e-06,
+      "loss": 2.8441,
+      "step": 246800
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.147031422340094e-06,
-      "loss": 2.6954,
-      "step": 228900
+      "learning_rate": 9.155787007933935e-06,
+      "loss": 2.8453,
+      "step": 246900
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.129183904394416e-06,
-      "loss": 2.6991,
-      "step": 229000
+      "learning_rate": 9.139244191817261e-06,
+      "loss": 2.8397,
+      "step": 247000
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.111336386448737e-06,
-      "loss": 2.7004,
-      "step": 229100
+      "learning_rate": 9.122701375700589e-06,
+      "loss": 2.8373,
+      "step": 247100
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.093488868503058e-06,
-      "loss": 2.6985,
-      "step": 229200
+      "learning_rate": 9.106158559583917e-06,
+      "loss": 2.8298,
+      "step": 247200
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.07564135055738e-06,
-      "loss": 2.7126,
-      "step": 229300
+      "learning_rate": 9.089615743467241e-06,
+      "loss": 2.8346,
+      "step": 247300
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.057793832611699e-06,
-      "loss": 2.7081,
-      "step": 229400
+      "learning_rate": 9.07307292735057e-06,
+      "loss": 2.8287,
+      "step": 247400
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.03994631466602e-06,
-      "loss": 2.6968,
-      "step": 229500
+      "learning_rate": 9.056530111233896e-06,
+      "loss": 2.8353,
+      "step": 247500
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.022098796720341e-06,
-      "loss": 2.6953,
-      "step": 229600
+      "learning_rate": 9.039987295117223e-06,
+      "loss": 2.8378,
+      "step": 247600
     },
     {
       "epoch": 0.82,
-      "learning_rate": 9.00425127877466e-06,
-      "loss": 2.7023,
-      "step": 229700
+      "learning_rate": 9.02344447900055e-06,
+      "loss": 2.8492,
+      "step": 247700
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.986403760828982e-06,
-      "loss": 2.6988,
-      "step": 229800
+      "learning_rate": 9.006901662883876e-06,
+      "loss": 2.8387,
+      "step": 247800
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.968556242883303e-06,
-      "loss": 2.7011,
-      "step": 229900
+      "learning_rate": 8.990358846767204e-06,
+      "loss": 2.8389,
+      "step": 247900
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.950708724937623e-06,
-      "loss": 2.7062,
-      "step": 230000
+      "learning_rate": 8.97381603065053e-06,
+      "loss": 2.8315,
+      "step": 248000
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.932861206991944e-06,
-      "loss": 2.6967,
-      "step": 230100
+      "learning_rate": 8.957273214533856e-06,
+      "loss": 2.8475,
+      "step": 248100
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.915013689046265e-06,
-      "loss": 2.6912,
-      "step": 230200
+      "learning_rate": 8.940730398417184e-06,
+      "loss": 2.8457,
+      "step": 248200
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.897166171100585e-06,
-      "loss": 2.7043,
-      "step": 230300
+      "learning_rate": 8.92418758230051e-06,
+      "loss": 2.8311,
+      "step": 248300
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.879318653154906e-06,
-      "loss": 2.6933,
-      "step": 230400
+      "learning_rate": 8.907644766183837e-06,
+      "loss": 2.8399,
+      "step": 248400
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.861471135209227e-06,
-      "loss": 2.7098,
-      "step": 230500
+      "learning_rate": 8.891101950067165e-06,
+      "loss": 2.8382,
+      "step": 248500
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.843623617263548e-06,
-      "loss": 2.6921,
-      "step": 230600
+      "learning_rate": 8.87455913395049e-06,
+      "loss": 2.8382,
+      "step": 248600
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.82577609931787e-06,
-      "loss": 2.6961,
-      "step": 230700
+      "learning_rate": 8.858016317833819e-06,
+      "loss": 2.8487,
+      "step": 248700
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.807928581372189e-06,
-      "loss": 2.6977,
-      "step": 230800
+      "learning_rate": 8.841473501717145e-06,
+      "loss": 2.8409,
+      "step": 248800
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.79008106342651e-06,
-      "loss": 2.7012,
-      "step": 230900
+      "learning_rate": 8.824930685600471e-06,
+      "loss": 2.8542,
+      "step": 248900
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.77223354548083e-06,
-      "loss": 2.7062,
-      "step": 231000
+      "learning_rate": 8.808387869483799e-06,
+      "loss": 2.833,
+      "step": 249000
     },
     {
       "epoch": 0.82,
-      "learning_rate": 8.754386027535151e-06,
-      "loss": 2.7066,
-      "step": 231100
+      "learning_rate": 8.791845053367125e-06,
+      "loss": 2.8466,
+      "step": 249100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.775302237250452e-06,
+      "loss": 2.8222,
+      "step": 249200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.75875942113378e-06,
+      "loss": 2.8427,
+      "step": 249300
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.736538509589472e-06,
-      "loss": 2.6948,
-      "step": 231200
+      "learning_rate": 8.742216605017106e-06,
+      "loss": 2.845,
+      "step": 249400
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.718690991643792e-06,
-      "loss": 2.7013,
-      "step": 231300
+      "learning_rate": 8.725673788900432e-06,
+      "loss": 2.8341,
+      "step": 249500
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.700843473698113e-06,
-      "loss": 2.7004,
-      "step": 231400
+      "learning_rate": 8.70913097278376e-06,
+      "loss": 2.8466,
+      "step": 249600
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.682995955752434e-06,
-      "loss": 2.6943,
-      "step": 231500
+      "learning_rate": 8.692588156667086e-06,
+      "loss": 2.836,
+      "step": 249700
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.665148437806754e-06,
-      "loss": 2.6959,
-      "step": 231600
+      "learning_rate": 8.676045340550414e-06,
+      "loss": 2.8352,
+      "step": 249800
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.647300919861075e-06,
-      "loss": 2.6993,
-      "step": 231700
+      "learning_rate": 8.659502524433739e-06,
+      "loss": 2.8256,
+      "step": 249900
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.629453401915396e-06,
-      "loss": 2.6986,
-      "step": 231800
+      "learning_rate": 8.642959708317066e-06,
+      "loss": 2.8283,
+      "step": 250000
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.611605883969717e-06,
-      "loss": 2.7002,
-      "step": 231900
+      "learning_rate": 8.626416892200394e-06,
+      "loss": 2.8248,
+      "step": 250100
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.593758366024039e-06,
-      "loss": 2.6978,
-      "step": 232000
+      "learning_rate": 8.60987407608372e-06,
+      "loss": 2.8272,
+      "step": 250200
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.575910848078358e-06,
-      "loss": 2.6906,
-      "step": 232100
+      "learning_rate": 8.593331259967047e-06,
+      "loss": 2.8367,
+      "step": 250300
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.55806333013268e-06,
-      "loss": 2.6979,
-      "step": 232200
+      "learning_rate": 8.576788443850375e-06,
+      "loss": 2.8394,
+      "step": 250400
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.540215812186999e-06,
-      "loss": 2.7052,
-      "step": 232300
+      "learning_rate": 8.560245627733701e-06,
+      "loss": 2.846,
+      "step": 250500
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.52236829424132e-06,
-      "loss": 2.6965,
-      "step": 232400
+      "learning_rate": 8.543702811617027e-06,
+      "loss": 2.8342,
+      "step": 250600
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.504520776295641e-06,
-      "loss": 2.6953,
-      "step": 232500
+      "learning_rate": 8.527159995500353e-06,
+      "loss": 2.8389,
+      "step": 250700
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.48667325834996e-06,
-      "loss": 2.6928,
-      "step": 232600
+      "learning_rate": 8.510617179383681e-06,
+      "loss": 2.8372,
+      "step": 250800
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.468825740404282e-06,
-      "loss": 2.7064,
-      "step": 232700
+      "learning_rate": 8.49407436326701e-06,
+      "loss": 2.8472,
+      "step": 250900
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.450978222458603e-06,
-      "loss": 2.6998,
-      "step": 232800
+      "learning_rate": 8.477531547150334e-06,
+      "loss": 2.8384,
+      "step": 251000
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.433130704512923e-06,
-      "loss": 2.6953,
-      "step": 232900
+      "learning_rate": 8.460988731033662e-06,
+      "loss": 2.8422,
+      "step": 251100
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.415283186567246e-06,
-      "loss": 2.691,
-      "step": 233000
+      "learning_rate": 8.444445914916988e-06,
+      "loss": 2.8478,
+      "step": 251200
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.397435668621565e-06,
-      "loss": 2.6897,
-      "step": 233100
+      "learning_rate": 8.427903098800316e-06,
+      "loss": 2.8366,
+      "step": 251300
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.379588150675886e-06,
-      "loss": 2.7072,
-      "step": 233200
+      "learning_rate": 8.411360282683642e-06,
+      "loss": 2.8453,
+      "step": 251400
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.361740632730208e-06,
-      "loss": 2.6961,
-      "step": 233300
+      "learning_rate": 8.394817466566968e-06,
+      "loss": 2.8398,
+      "step": 251500
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.343893114784527e-06,
-      "loss": 2.7006,
-      "step": 233400
+      "learning_rate": 8.378274650450296e-06,
+      "loss": 2.8468,
+      "step": 251600
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.326045596838848e-06,
-      "loss": 2.695,
-      "step": 233500
+      "learning_rate": 8.361731834333623e-06,
+      "loss": 2.8505,
+      "step": 251700
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.30819807889317e-06,
-      "loss": 2.6959,
-      "step": 233600
+      "learning_rate": 8.345189018216949e-06,
+      "loss": 2.8423,
+      "step": 251800
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.290350560947489e-06,
-      "loss": 2.685,
-      "step": 233700
+      "learning_rate": 8.328646202100277e-06,
+      "loss": 2.8399,
+      "step": 251900
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.27250304300181e-06,
-      "loss": 2.701,
-      "step": 233800
+      "learning_rate": 8.312103385983603e-06,
+      "loss": 2.842,
+      "step": 252000
     },
     {
       "epoch": 0.83,
-      "learning_rate": 8.25465552505613e-06,
-      "loss": 2.7057,
-      "step": 233900
+      "learning_rate": 8.29556056986693e-06,
+      "loss": 2.8289,
+      "step": 252100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.279017753750257e-06,
+      "loss": 2.8321,
+      "step": 252200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.262474937633583e-06,
+      "loss": 2.8468,
+      "step": 252300
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.236808007110451e-06,
-      "loss": 2.6963,
-      "step": 234000
+      "learning_rate": 8.245932121516911e-06,
+      "loss": 2.8482,
+      "step": 252400
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.218960489164772e-06,
-      "loss": 2.6934,
-      "step": 234100
+      "learning_rate": 8.229389305400237e-06,
+      "loss": 2.8364,
+      "step": 252500
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.201112971219092e-06,
-      "loss": 2.7008,
-      "step": 234200
+      "learning_rate": 8.212846489283564e-06,
+      "loss": 2.8432,
+      "step": 252600
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.183265453273415e-06,
-      "loss": 2.7019,
-      "step": 234300
+      "learning_rate": 8.196303673166892e-06,
+      "loss": 2.8301,
+      "step": 252700
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.165417935327734e-06,
-      "loss": 2.6915,
-      "step": 234400
+      "learning_rate": 8.179760857050218e-06,
+      "loss": 2.837,
+      "step": 252800
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.147570417382055e-06,
-      "loss": 2.702,
-      "step": 234500
+      "learning_rate": 8.163218040933544e-06,
+      "loss": 2.8518,
+      "step": 252900
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.129722899436377e-06,
-      "loss": 2.6931,
-      "step": 234600
+      "learning_rate": 8.146675224816872e-06,
+      "loss": 2.8275,
+      "step": 253000
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.111875381490696e-06,
-      "loss": 2.6943,
-      "step": 234700
+      "learning_rate": 8.130132408700198e-06,
+      "loss": 2.8246,
+      "step": 253100
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.094027863545017e-06,
-      "loss": 2.6886,
-      "step": 234800
+      "learning_rate": 8.113589592583524e-06,
+      "loss": 2.8418,
+      "step": 253200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.097046776466852e-06,
+      "loss": 2.8459,
+      "step": 253300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.080503960350179e-06,
+      "loss": 2.8349,
+      "step": 253400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.063961144233507e-06,
+      "loss": 2.8345,
+      "step": 253500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.047418328116833e-06,
+      "loss": 2.848,
+      "step": 253600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.030875512000159e-06,
+      "loss": 2.832,
+      "step": 253700
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.076180345599339e-06,
-      "loss": 2.6973,
-      "step": 234900
+      "learning_rate": 8.014332695883487e-06,
+      "loss": 2.8447,
+      "step": 253800
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.058332827653658e-06,
-      "loss": 2.6945,
-      "step": 235000
+      "learning_rate": 7.997789879766811e-06,
+      "loss": 2.8392,
+      "step": 253900
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.04048530970798e-06,
-      "loss": 2.6986,
-      "step": 235100
+      "learning_rate": 7.98124706365014e-06,
+      "loss": 2.8382,
+      "step": 254000
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.0226377917623e-06,
-      "loss": 2.6973,
-      "step": 235200
+      "learning_rate": 7.964704247533467e-06,
+      "loss": 2.8415,
+      "step": 254100
     },
     {
       "epoch": 0.84,
-      "learning_rate": 8.00479027381662e-06,
-      "loss": 2.6908,
-      "step": 235300
+      "learning_rate": 7.948161431416794e-06,
+      "loss": 2.8306,
+      "step": 254200
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.986942755870941e-06,
-      "loss": 2.6937,
-      "step": 235400
+      "learning_rate": 7.93161861530012e-06,
+      "loss": 2.8307,
+      "step": 254300
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.96909523792526e-06,
-      "loss": 2.7081,
-      "step": 235500
+      "learning_rate": 7.915075799183446e-06,
+      "loss": 2.8336,
+      "step": 254400
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.951247719979584e-06,
-      "loss": 2.6974,
-      "step": 235600
+      "learning_rate": 7.898532983066774e-06,
+      "loss": 2.8262,
+      "step": 254500
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.933400202033905e-06,
-      "loss": 2.6976,
-      "step": 235700
+      "learning_rate": 7.881990166950102e-06,
+      "loss": 2.8357,
+      "step": 254600
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.915552684088224e-06,
-      "loss": 2.6937,
-      "step": 235800
+      "learning_rate": 7.865447350833426e-06,
+      "loss": 2.8355,
+      "step": 254700
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.897705166142546e-06,
-      "loss": 2.6938,
-      "step": 235900
+      "learning_rate": 7.848904534716754e-06,
+      "loss": 2.8467,
+      "step": 254800
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.879857648196865e-06,
-      "loss": 2.6963,
-      "step": 236000
+      "learning_rate": 7.832361718600082e-06,
+      "loss": 2.8396,
+      "step": 254900
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.862010130251186e-06,
-      "loss": 2.6978,
-      "step": 236100
+      "learning_rate": 7.815818902483408e-06,
+      "loss": 2.8361,
+      "step": 255000
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.844162612305508e-06,
-      "loss": 2.6944,
-      "step": 236200
+      "learning_rate": 7.799276086366735e-06,
+      "loss": 2.836,
+      "step": 255100
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.826315094359827e-06,
-      "loss": 2.6979,
-      "step": 236300
+      "learning_rate": 7.782733270250061e-06,
+      "loss": 2.8461,
+      "step": 255200
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.808467576414148e-06,
-      "loss": 2.6917,
-      "step": 236400
+      "learning_rate": 7.766190454133389e-06,
+      "loss": 2.8397,
+      "step": 255300
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.79062005846847e-06,
-      "loss": 2.6964,
-      "step": 236500
+      "epoch": 0.85,
+      "learning_rate": 7.749647638016715e-06,
+      "loss": 2.8402,
+      "step": 255400
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.772772540522789e-06,
-      "loss": 2.6937,
-      "step": 236600
+      "epoch": 0.85,
+      "learning_rate": 7.733104821900041e-06,
+      "loss": 2.8321,
+      "step": 255500
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 7.75492502257711e-06,
-      "loss": 2.7025,
-      "step": 236700
+      "epoch": 0.85,
+      "learning_rate": 7.71656200578337e-06,
+      "loss": 2.8404,
+      "step": 255600
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.737077504631431e-06,
-      "loss": 2.6998,
-      "step": 236800
+      "learning_rate": 7.700019189666695e-06,
+      "loss": 2.8484,
+      "step": 255700
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.719229986685753e-06,
-      "loss": 2.6884,
-      "step": 236900
+      "learning_rate": 7.683476373550022e-06,
+      "loss": 2.8381,
+      "step": 255800
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.701382468740074e-06,
-      "loss": 2.7001,
-      "step": 237000
+      "learning_rate": 7.66693355743335e-06,
+      "loss": 2.8358,
+      "step": 255900
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.683534950794393e-06,
-      "loss": 2.6938,
-      "step": 237100
+      "learning_rate": 7.650390741316676e-06,
+      "loss": 2.834,
+      "step": 256000
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.665687432848715e-06,
-      "loss": 2.6937,
-      "step": 237200
+      "learning_rate": 7.633847925200004e-06,
+      "loss": 2.8357,
+      "step": 256100
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.647839914903036e-06,
-      "loss": 2.6936,
-      "step": 237300
+      "learning_rate": 7.617305109083329e-06,
+      "loss": 2.8387,
+      "step": 256200
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.629992396957355e-06,
-      "loss": 2.6893,
-      "step": 237400
+      "learning_rate": 7.600762292966657e-06,
+      "loss": 2.8461,
+      "step": 256300
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.6121448790116766e-06,
-      "loss": 2.687,
-      "step": 237500
+      "learning_rate": 7.584219476849984e-06,
+      "loss": 2.8485,
+      "step": 256400
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.594297361065997e-06,
-      "loss": 2.6948,
-      "step": 237600
+      "learning_rate": 7.5676766607333095e-06,
+      "loss": 2.8457,
+      "step": 256500
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.576449843120317e-06,
-      "loss": 2.6913,
-      "step": 237700
+      "learning_rate": 7.5511338446166366e-06,
+      "loss": 2.8417,
+      "step": 256600
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.558602325174638e-06,
-      "loss": 2.6908,
-      "step": 237800
+      "learning_rate": 7.5345910284999645e-06,
+      "loss": 2.8327,
+      "step": 256700
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.540754807228959e-06,
-      "loss": 2.6938,
-      "step": 237900
+      "learning_rate": 7.5180482123832916e-06,
+      "loss": 2.8432,
+      "step": 256800
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.522907289283279e-06,
-      "loss": 2.6906,
-      "step": 238000
+      "learning_rate": 7.501505396266617e-06,
+      "loss": 2.8293,
+      "step": 256900
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.505059771337601e-06,
-      "loss": 2.6924,
-      "step": 238100
+      "learning_rate": 7.484962580149944e-06,
+      "loss": 2.8262,
+      "step": 257000
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.487212253391922e-06,
-      "loss": 2.7051,
-      "step": 238200
+      "learning_rate": 7.468419764033271e-06,
+      "loss": 2.8373,
+      "step": 257100
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.469364735446242e-06,
-      "loss": 2.6969,
-      "step": 238300
+      "learning_rate": 7.451876947916599e-06,
+      "loss": 2.8446,
+      "step": 257200
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.451517217500563e-06,
-      "loss": 2.6976,
-      "step": 238400
+      "learning_rate": 7.435334131799924e-06,
+      "loss": 2.847,
+      "step": 257300
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.433669699554884e-06,
-      "loss": 2.6947,
-      "step": 238500
+      "learning_rate": 7.4187913156832515e-06,
+      "loss": 2.8559,
+      "step": 257400
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.415822181609204e-06,
-      "loss": 2.6945,
-      "step": 238600
+      "learning_rate": 7.4022484995665785e-06,
+      "loss": 2.8495,
+      "step": 257500
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.397974663663524e-06,
-      "loss": 2.6951,
-      "step": 238700
+      "learning_rate": 7.385705683449905e-06,
+      "loss": 2.8289,
+      "step": 257600
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.3801271457178456e-06,
-      "loss": 2.6876,
-      "step": 238800
+      "learning_rate": 7.369162867333232e-06,
+      "loss": 2.8264,
+      "step": 257700
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.362279627772166e-06,
-      "loss": 2.6975,
-      "step": 238900
+      "learning_rate": 7.352620051216559e-06,
+      "loss": 2.8355,
+      "step": 257800
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.344432109826486e-06,
-      "loss": 2.6884,
-      "step": 239000
+      "learning_rate": 7.336077235099886e-06,
+      "loss": 2.8328,
+      "step": 257900
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.326584591880807e-06,
-      "loss": 2.6911,
-      "step": 239100
+      "learning_rate": 7.319534418983212e-06,
+      "loss": 2.8332,
+      "step": 258000
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.308737073935128e-06,
-      "loss": 2.696,
-      "step": 239200
+      "learning_rate": 7.302991602866539e-06,
+      "loss": 2.8245,
+      "step": 258100
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.290889555989448e-06,
-      "loss": 2.6989,
-      "step": 239300
+      "learning_rate": 7.286448786749866e-06,
+      "loss": 2.8352,
+      "step": 258200
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.27304203804377e-06,
-      "loss": 2.6914,
-      "step": 239400
+      "learning_rate": 7.2699059706331935e-06,
+      "loss": 2.8484,
+      "step": 258300
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.255194520098091e-06,
-      "loss": 2.6928,
-      "step": 239500
+      "learning_rate": 7.25336315451652e-06,
+      "loss": 2.8606,
+      "step": 258400
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.237347002152411e-06,
-      "loss": 2.6887,
-      "step": 239600
+      "learning_rate": 7.236820338399847e-06,
+      "loss": 2.8424,
+      "step": 258500
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.219499484206732e-06,
-      "loss": 2.7037,
-      "step": 239700
+      "learning_rate": 7.220277522283174e-06,
+      "loss": 2.8418,
+      "step": 258600
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.201651966261053e-06,
-      "loss": 2.6903,
-      "step": 239800
+      "learning_rate": 7.203734706166501e-06,
+      "loss": 2.8362,
+      "step": 258700
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.183804448315373e-06,
-      "loss": 2.6958,
-      "step": 239900
+      "learning_rate": 7.187191890049827e-06,
+      "loss": 2.8514,
+      "step": 258800
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.165956930369694e-06,
-      "loss": 2.6902,
-      "step": 240000
+      "learning_rate": 7.170649073933154e-06,
+      "loss": 2.8463,
+      "step": 258900
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.148109412424015e-06,
-      "loss": 2.6947,
-      "step": 240100
+      "learning_rate": 7.154106257816481e-06,
+      "loss": 2.8246,
+      "step": 259000
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.130261894478335e-06,
-      "loss": 2.7103,
-      "step": 240200
+      "learning_rate": 7.1375634416998075e-06,
+      "loss": 2.8229,
+      "step": 259100
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.112414376532655e-06,
-      "loss": 2.7019,
-      "step": 240300
+      "learning_rate": 7.121020625583135e-06,
+      "loss": 2.8392,
+      "step": 259200
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.0945668585869765e-06,
-      "loss": 2.7102,
-      "step": 240400
+      "learning_rate": 7.104477809466462e-06,
+      "loss": 2.8228,
+      "step": 259300
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.076719340641297e-06,
-      "loss": 2.6914,
-      "step": 240500
+      "learning_rate": 7.087934993349789e-06,
+      "loss": 2.8432,
+      "step": 259400
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.058871822695617e-06,
-      "loss": 2.6956,
-      "step": 240600
+      "learning_rate": 7.071392177233115e-06,
+      "loss": 2.834,
+      "step": 259500
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.041024304749939e-06,
-      "loss": 2.6888,
-      "step": 240700
+      "learning_rate": 7.054849361116442e-06,
+      "loss": 2.8205,
+      "step": 259600
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.02317678680426e-06,
-      "loss": 2.6916,
-      "step": 240800
+      "learning_rate": 7.038306544999769e-06,
+      "loss": 2.8195,
+      "step": 259700
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.005329268858581e-06,
-      "loss": 2.6881,
-      "step": 240900
+      "learning_rate": 7.021763728883096e-06,
+      "loss": 2.8402,
+      "step": 259800
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.987481750912901e-06,
-      "loss": 2.6921,
-      "step": 241000
+      "learning_rate": 7.0052209127664224e-06,
+      "loss": 2.839,
+      "step": 259900
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.969634232967222e-06,
-      "loss": 2.7049,
-      "step": 241100
+      "learning_rate": 6.9886780966497495e-06,
+      "loss": 2.8289,
+      "step": 260000
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.951786715021542e-06,
-      "loss": 2.6928,
-      "step": 241200
+      "learning_rate": 6.972135280533077e-06,
+      "loss": 2.8263,
+      "step": 260100
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.933939197075863e-06,
-      "loss": 2.6975,
-      "step": 241300
+      "learning_rate": 6.955592464416402e-06,
+      "loss": 2.8468,
+      "step": 260200
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.916091679130184e-06,
-      "loss": 2.6956,
-      "step": 241400
+      "learning_rate": 6.939049648299729e-06,
+      "loss": 2.8396,
+      "step": 260300
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.898244161184504e-06,
-      "loss": 2.6812,
-      "step": 241500
+      "learning_rate": 6.922506832183057e-06,
+      "loss": 2.8385,
+      "step": 260400
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.880396643238825e-06,
-      "loss": 2.6887,
-      "step": 241600
+      "learning_rate": 6.905964016066384e-06,
+      "loss": 2.8326,
+      "step": 260500
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.8625491252931456e-06,
-      "loss": 2.6962,
-      "step": 241700
+      "learning_rate": 6.889421199949709e-06,
+      "loss": 2.8404,
+      "step": 260600
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.844701607347466e-06,
-      "loss": 2.689,
-      "step": 241800
+      "learning_rate": 6.8728783838330365e-06,
+      "loss": 2.8476,
+      "step": 260700
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.826854089401786e-06,
-      "loss": 2.689,
-      "step": 241900
+      "learning_rate": 6.856335567716364e-06,
+      "loss": 2.8282,
+      "step": 260800
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.809006571456108e-06,
-      "loss": 2.6999,
-      "step": 242000
+      "learning_rate": 6.8397927515996915e-06,
+      "loss": 2.8246,
+      "step": 260900
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.791159053510429e-06,
-      "loss": 2.6883,
-      "step": 242100
+      "learning_rate": 6.823249935483017e-06,
+      "loss": 2.8387,
+      "step": 261000
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.77331153556475e-06,
-      "loss": 2.6956,
-      "step": 242200
+      "learning_rate": 6.806707119366344e-06,
+      "loss": 2.8449,
+      "step": 261100
     },
     {
       "epoch": 0.86,
-      "learning_rate": 6.75546401761907e-06,
-      "loss": 2.7015,
-      "step": 242300
+      "learning_rate": 6.790164303249672e-06,
+      "loss": 2.8248,
+      "step": 261200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.773621487132997e-06,
+      "loss": 2.8474,
+      "step": 261300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.757078671016324e-06,
+      "loss": 2.8293,
+      "step": 261400
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.737616499673391e-06,
-      "loss": 2.688,
-      "step": 242400
+      "learning_rate": 6.740535854899651e-06,
+      "loss": 2.8398,
+      "step": 261500
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.719768981727712e-06,
-      "loss": 2.694,
-      "step": 242500
+      "learning_rate": 6.7239930387829785e-06,
+      "loss": 2.839,
+      "step": 261600
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.701921463782032e-06,
-      "loss": 2.688,
-      "step": 242600
+      "learning_rate": 6.707450222666305e-06,
+      "loss": 2.8478,
+      "step": 261700
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.684073945836353e-06,
-      "loss": 2.6889,
-      "step": 242700
+      "learning_rate": 6.690907406549632e-06,
+      "loss": 2.8224,
+      "step": 261800
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.666226427890673e-06,
-      "loss": 2.6904,
-      "step": 242800
+      "learning_rate": 6.674364590432959e-06,
+      "loss": 2.818,
+      "step": 261900
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.648378909944994e-06,
-      "loss": 2.6881,
-      "step": 242900
+      "learning_rate": 6.657821774316286e-06,
+      "loss": 2.8507,
+      "step": 262000
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.6305313919993146e-06,
-      "loss": 2.689,
-      "step": 243000
+      "learning_rate": 6.641278958199612e-06,
+      "loss": 2.8324,
+      "step": 262100
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.612683874053635e-06,
-      "loss": 2.6946,
-      "step": 243100
+      "learning_rate": 6.624736142082939e-06,
+      "loss": 2.8242,
+      "step": 262200
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.594836356107955e-06,
-      "loss": 2.6888,
-      "step": 243200
+      "learning_rate": 6.608193325966266e-06,
+      "loss": 2.8304,
+      "step": 262300
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.576988838162277e-06,
-      "loss": 2.6866,
-      "step": 243300
+      "learning_rate": 6.591650509849593e-06,
+      "loss": 2.8319,
+      "step": 262400
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.559141320216599e-06,
-      "loss": 2.7064,
-      "step": 243400
+      "learning_rate": 6.57510769373292e-06,
+      "loss": 2.8367,
+      "step": 262500
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.541293802270919e-06,
-      "loss": 2.6871,
-      "step": 243500
+      "learning_rate": 6.558564877616247e-06,
+      "loss": 2.8416,
+      "step": 262600
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.523446284325239e-06,
-      "loss": 2.6867,
-      "step": 243600
+      "learning_rate": 6.542022061499574e-06,
+      "loss": 2.8527,
+      "step": 262700
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.50559876637956e-06,
-      "loss": 2.6865,
-      "step": 243700
+      "learning_rate": 6.5254792453829e-06,
+      "loss": 2.8451,
+      "step": 262800
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.487751248433881e-06,
-      "loss": 2.6934,
-      "step": 243800
+      "learning_rate": 6.508936429266227e-06,
+      "loss": 2.814,
+      "step": 262900
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.469903730488201e-06,
-      "loss": 2.6892,
-      "step": 243900
+      "learning_rate": 6.492393613149554e-06,
+      "loss": 2.834,
+      "step": 263000
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.452056212542522e-06,
-      "loss": 2.7051,
-      "step": 244000
+      "learning_rate": 6.475850797032881e-06,
+      "loss": 2.8395,
+      "step": 263100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4593079809162075e-06,
+      "loss": 2.835,
+      "step": 263200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.4427651647995345e-06,
+      "loss": 2.8362,
+      "step": 263300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.426222348682862e-06,
+      "loss": 2.8301,
+      "step": 263400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 6.409679532566189e-06,
+      "loss": 2.8373,
+      "step": 263500
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.434208694596843e-06,
-      "loss": 2.6944,
-      "step": 244100
+      "learning_rate": 6.393136716449515e-06,
+      "loss": 2.8454,
+      "step": 263600
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.416361176651163e-06,
-      "loss": 2.7003,
-      "step": 244200
+      "learning_rate": 6.376593900332842e-06,
+      "loss": 2.8475,
+      "step": 263700
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.398513658705484e-06,
-      "loss": 2.6952,
-      "step": 244300
+      "learning_rate": 6.360051084216169e-06,
+      "loss": 2.8436,
+      "step": 263800
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.380666140759804e-06,
-      "loss": 2.6977,
-      "step": 244400
+      "learning_rate": 6.3435082680994944e-06,
+      "loss": 2.827,
+      "step": 263900
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.362818622814126e-06,
-      "loss": 2.6896,
-      "step": 244500
+      "learning_rate": 6.326965451982822e-06,
+      "loss": 2.834,
+      "step": 264000
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.344971104868447e-06,
-      "loss": 2.6967,
-      "step": 244600
+      "learning_rate": 6.3104226358661494e-06,
+      "loss": 2.8361,
+      "step": 264100
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.327123586922768e-06,
-      "loss": 2.6798,
-      "step": 244700
+      "learning_rate": 6.2938798197494765e-06,
+      "loss": 2.8224,
+      "step": 264200
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.309276068977088e-06,
-      "loss": 2.6945,
-      "step": 244800
+      "learning_rate": 6.277337003632802e-06,
+      "loss": 2.8473,
+      "step": 264300
     },
     {
       "epoch": 0.87,
-      "learning_rate": 6.291428551031408e-06,
-      "loss": 2.6842,
-      "step": 244900
+      "learning_rate": 6.260794187516129e-06,
+      "loss": 2.8312,
+      "step": 264400
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 6.2735810330857296e-06,
-      "loss": 2.6888,
-      "step": 245000
+      "epoch": 0.88,
+      "learning_rate": 6.244251371399457e-06,
+      "loss": 2.8242,
+      "step": 264500
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 6.25573351514005e-06,
-      "loss": 2.6997,
-      "step": 245100
+      "epoch": 0.88,
+      "learning_rate": 6.227708555282783e-06,
+      "loss": 2.8276,
+      "step": 264600
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.23788599719437e-06,
-      "loss": 2.6885,
-      "step": 245200
+      "learning_rate": 6.21116573916611e-06,
+      "loss": 2.8432,
+      "step": 264700
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.220038479248691e-06,
-      "loss": 2.7027,
-      "step": 245300
+      "learning_rate": 6.1946229230494364e-06,
+      "loss": 2.8509,
+      "step": 264800
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.202190961303012e-06,
-      "loss": 2.6942,
-      "step": 245400
+      "learning_rate": 6.178080106932764e-06,
+      "loss": 2.8381,
+      "step": 264900
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.184343443357333e-06,
-      "loss": 2.6893,
-      "step": 245500
+      "learning_rate": 6.161537290816091e-06,
+      "loss": 2.8419,
+      "step": 265000
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.1664959254116535e-06,
-      "loss": 2.7014,
-      "step": 245600
+      "learning_rate": 6.144994474699417e-06,
+      "loss": 2.8342,
+      "step": 265100
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.148648407465974e-06,
-      "loss": 2.6891,
-      "step": 245700
+      "learning_rate": 6.128451658582744e-06,
+      "loss": 2.8224,
+      "step": 265200
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.130800889520295e-06,
-      "loss": 2.702,
-      "step": 245800
+      "learning_rate": 6.111908842466071e-06,
+      "loss": 2.8375,
+      "step": 265300
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.112953371574615e-06,
-      "loss": 2.6909,
-      "step": 245900
+      "learning_rate": 6.095366026349398e-06,
+      "loss": 2.8436,
+      "step": 265400
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.095105853628936e-06,
-      "loss": 2.6906,
-      "step": 246000
+      "learning_rate": 6.078823210232724e-06,
+      "loss": 2.8336,
+      "step": 265500
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.077258335683257e-06,
-      "loss": 2.6964,
-      "step": 246100
+      "learning_rate": 6.062280394116051e-06,
+      "loss": 2.8403,
+      "step": 265600
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.059410817737577e-06,
-      "loss": 2.7003,
-      "step": 246200
+      "learning_rate": 6.045737577999378e-06,
+      "loss": 2.841,
+      "step": 265700
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.0415632997918986e-06,
-      "loss": 2.6929,
-      "step": 246300
+      "learning_rate": 6.0291947618827055e-06,
+      "loss": 2.8326,
+      "step": 265800
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.023715781846219e-06,
-      "loss": 2.6888,
-      "step": 246400
+      "learning_rate": 6.012651945766032e-06,
+      "loss": 2.8464,
+      "step": 265900
     },
     {
       "epoch": 0.88,
-      "learning_rate": 6.005868263900539e-06,
-      "loss": 2.701,
-      "step": 246500
+      "learning_rate": 5.996109129649359e-06,
+      "loss": 2.829,
+      "step": 266000
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.9880207459548605e-06,
-      "loss": 2.6993,
-      "step": 246600
+      "learning_rate": 5.979566313532686e-06,
+      "loss": 2.8414,
+      "step": 266100
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.970173228009181e-06,
-      "loss": 2.6882,
-      "step": 246700
+      "learning_rate": 5.963023497416012e-06,
+      "loss": 2.8409,
+      "step": 266200
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.952325710063502e-06,
-      "loss": 2.6969,
-      "step": 246800
+      "learning_rate": 5.946480681299339e-06,
+      "loss": 2.8403,
+      "step": 266300
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.9344781921178225e-06,
-      "loss": 2.7008,
-      "step": 246900
+      "learning_rate": 5.929937865182665e-06,
+      "loss": 2.8272,
+      "step": 266400
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.916630674172143e-06,
-      "loss": 2.6964,
-      "step": 247000
+      "learning_rate": 5.913395049065993e-06,
+      "loss": 2.8372,
+      "step": 266500
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.898783156226464e-06,
-      "loss": 2.6921,
-      "step": 247100
+      "learning_rate": 5.8968522329493196e-06,
+      "loss": 2.8304,
+      "step": 266600
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.8809356382807844e-06,
-      "loss": 2.6875,
-      "step": 247200
+      "learning_rate": 5.880309416832647e-06,
+      "loss": 2.8216,
+      "step": 266700
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.863088120335105e-06,
-      "loss": 2.6913,
-      "step": 247300
+      "learning_rate": 5.863766600715973e-06,
+      "loss": 2.8344,
+      "step": 266800
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.845240602389426e-06,
-      "loss": 2.7004,
-      "step": 247400
+      "learning_rate": 5.847223784599301e-06,
+      "loss": 2.8193,
+      "step": 266900
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.827393084443747e-06,
-      "loss": 2.6947,
-      "step": 247500
+      "learning_rate": 5.830680968482627e-06,
+      "loss": 2.8471,
+      "step": 267000
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.809545566498068e-06,
-      "loss": 2.6968,
-      "step": 247600
+      "learning_rate": 5.814138152365954e-06,
+      "loss": 2.8366,
+      "step": 267100
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.791698048552388e-06,
-      "loss": 2.6929,
-      "step": 247700
+      "learning_rate": 5.79759533624928e-06,
+      "loss": 2.8332,
+      "step": 267200
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.773850530606708e-06,
-      "loss": 2.686,
-      "step": 247800
+      "learning_rate": 5.781052520132607e-06,
+      "loss": 2.8337,
+      "step": 267300
     },
     {
       "epoch": 0.88,
-      "learning_rate": 5.7560030126610295e-06,
-      "loss": 2.6899,
-      "step": 247900
+      "learning_rate": 5.7645097040159345e-06,
+      "loss": 2.8402,
+      "step": 267400
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.738155494715351e-06,
-      "loss": 2.6949,
-      "step": 248000
+      "learning_rate": 5.747966887899261e-06,
+      "loss": 2.8325,
+      "step": 267500
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.720307976769671e-06,
-      "loss": 2.6969,
-      "step": 248100
+      "learning_rate": 5.731424071782588e-06,
+      "loss": 2.8273,
+      "step": 267600
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.7024604588239915e-06,
-      "loss": 2.691,
-      "step": 248200
+      "learning_rate": 5.714881255665915e-06,
+      "loss": 2.8323,
+      "step": 267700
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.684612940878313e-06,
-      "loss": 2.6885,
-      "step": 248300
+      "learning_rate": 5.698338439549242e-06,
+      "loss": 2.8453,
+      "step": 267800
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.666765422932633e-06,
-      "loss": 2.69,
-      "step": 248400
+      "learning_rate": 5.681795623432568e-06,
+      "loss": 2.8241,
+      "step": 267900
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.6489179049869534e-06,
-      "loss": 2.6909,
-      "step": 248500
+      "learning_rate": 5.665252807315895e-06,
+      "loss": 2.8517,
+      "step": 268000
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.631070387041274e-06,
-      "loss": 2.7055,
-      "step": 248600
+      "learning_rate": 5.648709991199222e-06,
+      "loss": 2.8418,
+      "step": 268100
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.613222869095595e-06,
-      "loss": 2.6838,
-      "step": 248700
+      "learning_rate": 5.632167175082549e-06,
+      "loss": 2.8337,
+      "step": 268200
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.595375351149916e-06,
-      "loss": 2.6962,
-      "step": 248800
+      "learning_rate": 5.615624358965876e-06,
+      "loss": 2.8357,
+      "step": 268300
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.577527833204237e-06,
-      "loss": 2.6954,
-      "step": 248900
+      "learning_rate": 5.599081542849203e-06,
+      "loss": 2.8355,
+      "step": 268400
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.559680315258557e-06,
-      "loss": 2.6933,
-      "step": 249000
+      "learning_rate": 5.58253872673253e-06,
+      "loss": 2.8355,
+      "step": 268500
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.541832797312878e-06,
-      "loss": 2.6939,
-      "step": 249100
+      "learning_rate": 5.565995910615857e-06,
+      "loss": 2.8544,
+      "step": 268600
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.5239852793671986e-06,
-      "loss": 2.6894,
-      "step": 249200
+      "learning_rate": 5.549453094499183e-06,
+      "loss": 2.8397,
+      "step": 268700
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.50613776142152e-06,
-      "loss": 2.6872,
-      "step": 249300
+      "learning_rate": 5.532910278382509e-06,
+      "loss": 2.8375,
+      "step": 268800
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.48829024347584e-06,
-      "loss": 2.6983,
-      "step": 249400
+      "learning_rate": 5.516367462265836e-06,
+      "loss": 2.8249,
+      "step": 268900
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.4704427255301605e-06,
-      "loss": 2.6958,
-      "step": 249500
+      "learning_rate": 5.4998246461491634e-06,
+      "loss": 2.8402,
+      "step": 269000
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.452595207584482e-06,
-      "loss": 2.6866,
-      "step": 249600
+      "learning_rate": 5.4832818300324905e-06,
+      "loss": 2.824,
+      "step": 269100
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.434747689638802e-06,
-      "loss": 2.6886,
-      "step": 249700
+      "learning_rate": 5.466739013915817e-06,
+      "loss": 2.836,
+      "step": 269200
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.4169001716931225e-06,
-      "loss": 2.6916,
-      "step": 249800
+      "learning_rate": 5.450196197799144e-06,
+      "loss": 2.8463,
+      "step": 269300
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.399052653747444e-06,
-      "loss": 2.6907,
-      "step": 249900
+      "learning_rate": 5.433653381682471e-06,
+      "loss": 2.8218,
+      "step": 269400
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.381205135801765e-06,
-      "loss": 2.6826,
-      "step": 250000
+      "learning_rate": 5.417110565565798e-06,
+      "loss": 2.8284,
+      "step": 269500
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.363357617856085e-06,
-      "loss": 2.6888,
-      "step": 250100
+      "learning_rate": 5.400567749449124e-06,
+      "loss": 2.8436,
+      "step": 269600
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.345510099910406e-06,
-      "loss": 2.6791,
-      "step": 250200
+      "learning_rate": 5.384024933332451e-06,
+      "loss": 2.8317,
+      "step": 269700
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.327662581964726e-06,
-      "loss": 2.6788,
-      "step": 250300
+      "learning_rate": 5.367482117215778e-06,
+      "loss": 2.8338,
+      "step": 269800
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.309815064019047e-06,
-      "loss": 2.7038,
-      "step": 250400
+      "learning_rate": 5.350939301099105e-06,
+      "loss": 2.8463,
+      "step": 269900
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.2919675460733676e-06,
-      "loss": 2.6972,
-      "step": 250500
+      "learning_rate": 5.334396484982432e-06,
+      "loss": 2.8414,
+      "step": 270000
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.274120028127689e-06,
-      "loss": 2.6859,
-      "step": 250600
+      "learning_rate": 5.317853668865759e-06,
+      "loss": 2.822,
+      "step": 270100
     },
     {
       "epoch": 0.89,
-      "learning_rate": 5.256272510182009e-06,
-      "loss": 2.6911,
-      "step": 250700
+      "learning_rate": 5.301310852749086e-06,
+      "loss": 2.8352,
+      "step": 270200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.284768036632412e-06,
+      "loss": 2.8344,
+      "step": 270300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.268225220515739e-06,
+      "loss": 2.8375,
+      "step": 270400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.251682404399065e-06,
+      "loss": 2.8363,
+      "step": 270500
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.23842499223633e-06,
-      "loss": 2.6862,
-      "step": 250800
+      "learning_rate": 5.235139588282393e-06,
+      "loss": 2.8403,
+      "step": 270600
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.220577474290651e-06,
-      "loss": 2.6866,
-      "step": 250900
+      "learning_rate": 5.2185967721657195e-06,
+      "loss": 2.8441,
+      "step": 270700
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.202729956344971e-06,
-      "loss": 2.676,
-      "step": 251000
+      "learning_rate": 5.2020539560490466e-06,
+      "loss": 2.8146,
+      "step": 270800
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.1848824383992915e-06,
-      "loss": 2.6817,
-      "step": 251100
+      "learning_rate": 5.185511139932373e-06,
+      "loss": 2.8347,
+      "step": 270900
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.167034920453613e-06,
-      "loss": 2.6998,
-      "step": 251200
+      "learning_rate": 5.168968323815701e-06,
+      "loss": 2.8322,
+      "step": 271000
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.149187402507934e-06,
-      "loss": 2.6946,
-      "step": 251300
+      "learning_rate": 5.152425507699027e-06,
+      "loss": 2.8388,
+      "step": 271100
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.131339884562254e-06,
-      "loss": 2.6848,
-      "step": 251400
+      "learning_rate": 5.135882691582353e-06,
+      "loss": 2.8277,
+      "step": 271200
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.113492366616575e-06,
-      "loss": 2.7021,
-      "step": 251500
+      "learning_rate": 5.11933987546568e-06,
+      "loss": 2.8545,
+      "step": 271300
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.095644848670896e-06,
-      "loss": 2.695,
-      "step": 251600
+      "learning_rate": 5.102797059349007e-06,
+      "loss": 2.8294,
+      "step": 271400
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.077797330725216e-06,
-      "loss": 2.6968,
-      "step": 251700
+      "learning_rate": 5.086254243232334e-06,
+      "loss": 2.8369,
+      "step": 271500
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.059949812779537e-06,
-      "loss": 2.6969,
-      "step": 251800
+      "learning_rate": 5.069711427115661e-06,
+      "loss": 2.8506,
+      "step": 271600
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.042102294833858e-06,
-      "loss": 2.6867,
-      "step": 251900
+      "learning_rate": 5.053168610998988e-06,
+      "loss": 2.8408,
+      "step": 271700
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.024254776888178e-06,
-      "loss": 2.6927,
-      "step": 252000
+      "learning_rate": 5.036625794882315e-06,
+      "loss": 2.836,
+      "step": 271800
     },
     {
       "epoch": 0.9,
-      "learning_rate": 5.006407258942499e-06,
-      "loss": 2.692,
-      "step": 252100
+      "learning_rate": 5.020082978765642e-06,
+      "loss": 2.8291,
+      "step": 271900
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.98855974099682e-06,
-      "loss": 2.6792,
-      "step": 252200
+      "learning_rate": 5.003540162648968e-06,
+      "loss": 2.8429,
+      "step": 272000
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.97071222305114e-06,
-      "loss": 2.6827,
-      "step": 252300
+      "learning_rate": 4.986997346532295e-06,
+      "loss": 2.8281,
+      "step": 272100
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.952864705105461e-06,
-      "loss": 2.6947,
-      "step": 252400
+      "learning_rate": 4.970454530415622e-06,
+      "loss": 2.8337,
+      "step": 272200
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.9350171871597826e-06,
-      "loss": 2.7035,
-      "step": 252500
+      "learning_rate": 4.9539117142989485e-06,
+      "loss": 2.832,
+      "step": 272300
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.917169669214103e-06,
-      "loss": 2.6853,
-      "step": 252600
+      "learning_rate": 4.9373688981822755e-06,
+      "loss": 2.8279,
+      "step": 272400
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.899322151268423e-06,
-      "loss": 2.689,
-      "step": 252700
+      "learning_rate": 4.920826082065602e-06,
+      "loss": 2.8418,
+      "step": 272500
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.881474633322744e-06,
-      "loss": 2.6917,
-      "step": 252800
+      "learning_rate": 4.90428326594893e-06,
+      "loss": 2.8311,
+      "step": 272600
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.863627115377065e-06,
-      "loss": 2.6907,
-      "step": 252900
+      "learning_rate": 4.887740449832256e-06,
+      "loss": 2.8239,
+      "step": 272700
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.845779597431385e-06,
-      "loss": 2.6849,
-      "step": 253000
+      "learning_rate": 4.871197633715583e-06,
+      "loss": 2.8235,
+      "step": 272800
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.8279320794857065e-06,
-      "loss": 2.6868,
-      "step": 253100
+      "learning_rate": 4.854654817598909e-06,
+      "loss": 2.8388,
+      "step": 272900
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.810084561540027e-06,
-      "loss": 2.6929,
-      "step": 253200
+      "learning_rate": 4.838112001482237e-06,
+      "loss": 2.8403,
+      "step": 273000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821569185365563e-06,
+      "loss": 2.8258,
+      "step": 273100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8050263692488905e-06,
+      "loss": 2.8286,
+      "step": 273200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788483553132217e-06,
+      "loss": 2.8339,
+      "step": 273300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.771940737015544e-06,
+      "loss": 2.8328,
+      "step": 273400
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.792237043594348e-06,
-      "loss": 2.6968,
-      "step": 253300
+      "learning_rate": 4.755397920898871e-06,
+      "loss": 2.8372,
+      "step": 273500
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.774389525648668e-06,
-      "loss": 2.683,
-      "step": 253400
+      "epoch": 0.91,
+      "learning_rate": 4.738855104782197e-06,
+      "loss": 2.8434,
+      "step": 273600
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.756542007702989e-06,
-      "loss": 2.6751,
-      "step": 253500
+      "epoch": 0.91,
+      "learning_rate": 4.722312288665524e-06,
+      "loss": 2.8325,
+      "step": 273700
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.738694489757309e-06,
-      "loss": 2.6998,
-      "step": 253600
+      "learning_rate": 4.705769472548851e-06,
+      "loss": 2.8174,
+      "step": 273800
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.72084697181163e-06,
-      "loss": 2.7031,
-      "step": 253700
+      "learning_rate": 4.689226656432178e-06,
+      "loss": 2.8223,
+      "step": 273900
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.7029994538659516e-06,
-      "loss": 2.6882,
-      "step": 253800
+      "learning_rate": 4.6726838403155045e-06,
+      "loss": 2.8358,
+      "step": 274000
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.685151935920272e-06,
-      "loss": 2.6885,
-      "step": 253900
+      "learning_rate": 4.656141024198832e-06,
+      "loss": 2.8336,
+      "step": 274100
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.667304417974592e-06,
-      "loss": 2.6818,
-      "step": 254000
+      "learning_rate": 4.639598208082159e-06,
+      "loss": 2.8315,
+      "step": 274200
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.6494569000289135e-06,
-      "loss": 2.6897,
-      "step": 254100
+      "learning_rate": 4.623055391965486e-06,
+      "loss": 2.8413,
+      "step": 274300
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.631609382083234e-06,
-      "loss": 2.6905,
-      "step": 254200
+      "learning_rate": 4.606512575848812e-06,
+      "loss": 2.8361,
+      "step": 274400
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.613761864137554e-06,
-      "loss": 2.6938,
-      "step": 254300
+      "learning_rate": 4.589969759732139e-06,
+      "loss": 2.8091,
+      "step": 274500
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.5959143461918755e-06,
-      "loss": 2.6965,
-      "step": 254400
+      "learning_rate": 4.573426943615466e-06,
+      "loss": 2.8356,
+      "step": 274600
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.578066828246196e-06,
-      "loss": 2.6999,
-      "step": 254500
+      "learning_rate": 4.556884127498793e-06,
+      "loss": 2.8385,
+      "step": 274700
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.560219310300517e-06,
-      "loss": 2.6905,
-      "step": 254600
+      "learning_rate": 4.5403413113821194e-06,
+      "loss": 2.8336,
+      "step": 274800
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.5423717923548374e-06,
-      "loss": 2.6768,
-      "step": 254700
+      "learning_rate": 4.523798495265446e-06,
+      "loss": 2.8317,
+      "step": 274900
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.524524274409158e-06,
-      "loss": 2.6879,
-      "step": 254800
+      "learning_rate": 4.507255679148773e-06,
+      "loss": 2.8453,
+      "step": 275000
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.506676756463479e-06,
-      "loss": 2.6928,
-      "step": 254900
+      "learning_rate": 4.4907128630321e-06,
+      "loss": 2.835,
+      "step": 275100
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.488829238517799e-06,
-      "loss": 2.6959,
-      "step": 255000
+      "learning_rate": 4.474170046915427e-06,
+      "loss": 2.8117,
+      "step": 275200
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.470981720572121e-06,
-      "loss": 2.6813,
-      "step": 255100
+      "learning_rate": 4.457627230798753e-06,
+      "loss": 2.8333,
+      "step": 275300
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.453134202626441e-06,
-      "loss": 2.6957,
-      "step": 255200
+      "learning_rate": 4.44108441468208e-06,
+      "loss": 2.8303,
+      "step": 275400
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.435286684680761e-06,
-      "loss": 2.6822,
-      "step": 255300
+      "learning_rate": 4.424541598565407e-06,
+      "loss": 2.8325,
+      "step": 275500
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.4174391667350825e-06,
-      "loss": 2.695,
-      "step": 255400
+      "learning_rate": 4.407998782448734e-06,
+      "loss": 2.8383,
+      "step": 275600
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.399591648789403e-06,
-      "loss": 2.6949,
-      "step": 255500
+      "learning_rate": 4.3914559663320606e-06,
+      "loss": 2.8291,
+      "step": 275700
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.381744130843723e-06,
-      "loss": 2.6821,
-      "step": 255600
+      "learning_rate": 4.374913150215388e-06,
+      "loss": 2.8302,
+      "step": 275800
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.3638966128980445e-06,
-      "loss": 2.6904,
-      "step": 255700
+      "learning_rate": 4.358370334098715e-06,
+      "loss": 2.8235,
+      "step": 275900
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.346049094952366e-06,
-      "loss": 2.6912,
-      "step": 255800
+      "learning_rate": 4.341827517982041e-06,
+      "loss": 2.8427,
+      "step": 276000
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.328201577006686e-06,
-      "loss": 2.6931,
-      "step": 255900
+      "learning_rate": 4.325284701865368e-06,
+      "loss": 2.8216,
+      "step": 276100
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.3103540590610064e-06,
-      "loss": 2.686,
-      "step": 256000
+      "learning_rate": 4.308741885748695e-06,
+      "loss": 2.8482,
+      "step": 276200
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.292506541115327e-06,
-      "loss": 2.6957,
-      "step": 256100
+      "learning_rate": 4.292199069632022e-06,
+      "loss": 2.8404,
+      "step": 276300
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.274659023169648e-06,
-      "loss": 2.6824,
-      "step": 256200
+      "learning_rate": 4.275656253515348e-06,
+      "loss": 2.8344,
+      "step": 276400
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.256811505223969e-06,
-      "loss": 2.6841,
-      "step": 256300
+      "learning_rate": 4.2591134373986755e-06,
+      "loss": 2.8283,
+      "step": 276500
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.23896398727829e-06,
-      "loss": 2.6876,
-      "step": 256400
+      "learning_rate": 4.242570621282002e-06,
+      "loss": 2.8369,
+      "step": 276600
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.22111646933261e-06,
-      "loss": 2.6932,
-      "step": 256500
+      "learning_rate": 4.22602780516533e-06,
+      "loss": 2.8249,
+      "step": 276700
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.203268951386931e-06,
-      "loss": 2.6916,
-      "step": 256600
+      "learning_rate": 4.209484989048656e-06,
+      "loss": 2.8257,
+      "step": 276800
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.1854214334412516e-06,
-      "loss": 2.6874,
-      "step": 256700
+      "learning_rate": 4.192942172931983e-06,
+      "loss": 2.8261,
+      "step": 276900
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.167573915495572e-06,
-      "loss": 2.6823,
-      "step": 256800
+      "learning_rate": 4.176399356815309e-06,
+      "loss": 2.8238,
+      "step": 277000
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.149726397549892e-06,
-      "loss": 2.687,
-      "step": 256900
+      "learning_rate": 4.159856540698637e-06,
+      "loss": 2.8322,
+      "step": 277100
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.1318788796042135e-06,
-      "loss": 2.6951,
-      "step": 257000
+      "learning_rate": 4.143313724581963e-06,
+      "loss": 2.8356,
+      "step": 277200
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.114031361658535e-06,
-      "loss": 2.6977,
-      "step": 257100
+      "learning_rate": 4.1267709084652896e-06,
+      "loss": 2.8243,
+      "step": 277300
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.096183843712855e-06,
-      "loss": 2.6904,
-      "step": 257200
+      "learning_rate": 4.110228092348617e-06,
+      "loss": 2.8282,
+      "step": 277400
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.0783363257671755e-06,
-      "loss": 2.6816,
-      "step": 257300
+      "learning_rate": 4.093685276231944e-06,
+      "loss": 2.8338,
+      "step": 277500
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.060488807821497e-06,
-      "loss": 2.6873,
-      "step": 257400
+      "learning_rate": 4.077142460115271e-06,
+      "loss": 2.8397,
+      "step": 277600
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.042641289875817e-06,
-      "loss": 2.6894,
-      "step": 257500
+      "learning_rate": 4.060599643998597e-06,
+      "loss": 2.8358,
+      "step": 277700
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.024793771930138e-06,
-      "loss": 2.6893,
-      "step": 257600
+      "learning_rate": 4.044056827881924e-06,
+      "loss": 2.8282,
+      "step": 277800
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.006946253984459e-06,
-      "loss": 2.6965,
-      "step": 257700
+      "learning_rate": 4.027514011765251e-06,
+      "loss": 2.8238,
+      "step": 277900
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.989098736038779e-06,
-      "loss": 2.685,
-      "step": 257800
+      "learning_rate": 4.010971195648578e-06,
+      "loss": 2.8451,
+      "step": 278000
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.9712512180931e-06,
-      "loss": 2.6865,
-      "step": 257900
+      "learning_rate": 3.9944283795319045e-06,
+      "loss": 2.8329,
+      "step": 278100
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.953403700147421e-06,
-      "loss": 2.6911,
-      "step": 258000
+      "learning_rate": 3.9778855634152315e-06,
+      "loss": 2.8147,
+      "step": 278200
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.935556182201741e-06,
-      "loss": 2.6885,
-      "step": 258100
+      "learning_rate": 3.961342747298559e-06,
+      "loss": 2.821,
+      "step": 278300
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.917708664256062e-06,
-      "loss": 2.6951,
-      "step": 258200
+      "learning_rate": 3.944799931181886e-06,
+      "loss": 2.8466,
+      "step": 278400
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.899861146310383e-06,
-      "loss": 2.6964,
-      "step": 258300
+      "learning_rate": 3.928257115065212e-06,
+      "loss": 2.8231,
+      "step": 278500
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.882013628364704e-06,
-      "loss": 2.6863,
-      "step": 258400
+      "learning_rate": 3.911714298948538e-06,
+      "loss": 2.8435,
+      "step": 278600
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.864166110419024e-06,
-      "loss": 2.6828,
-      "step": 258500
+      "learning_rate": 3.895171482831866e-06,
+      "loss": 2.8269,
+      "step": 278700
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.8463185924733445e-06,
-      "loss": 2.6844,
-      "step": 258600
+      "learning_rate": 3.878628666715192e-06,
+      "loss": 2.8395,
+      "step": 278800
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.828471074527666e-06,
-      "loss": 2.6863,
-      "step": 258700
+      "learning_rate": 3.862085850598519e-06,
+      "loss": 2.8496,
+      "step": 278900
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.810623556581986e-06,
-      "loss": 2.6983,
-      "step": 258800
+      "learning_rate": 3.845543034481846e-06,
+      "loss": 2.8434,
+      "step": 279000
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.7927760386363073e-06,
-      "loss": 2.6873,
-      "step": 258900
+      "learning_rate": 3.8290002183651735e-06,
+      "loss": 2.8309,
+      "step": 279100
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.774928520690628e-06,
-      "loss": 2.6917,
-      "step": 259000
+      "learning_rate": 3.8124574022484998e-06,
+      "loss": 2.8288,
+      "step": 279200
     },
     {
       "epoch": 0.92,
-      "learning_rate": 3.7570810027449484e-06,
-      "loss": 2.6926,
-      "step": 259100
+      "learning_rate": 3.795914586131827e-06,
+      "loss": 2.8263,
+      "step": 279300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7793717700151535e-06,
+      "loss": 2.8135,
+      "step": 279400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7628289538984806e-06,
+      "loss": 2.8372,
+      "step": 279500
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.7392334847992692e-06,
-      "loss": 2.688,
-      "step": 259200
+      "learning_rate": 3.746286137781807e-06,
+      "loss": 2.826,
+      "step": 279600
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.7213859668535896e-06,
-      "loss": 2.6805,
-      "step": 259300
+      "learning_rate": 3.7297433216651334e-06,
+      "loss": 2.8173,
+      "step": 279700
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.7035384489079104e-06,
-      "loss": 2.6856,
-      "step": 259400
+      "learning_rate": 3.713200505548461e-06,
+      "loss": 2.8202,
+      "step": 279800
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.6856909309622316e-06,
-      "loss": 2.6907,
-      "step": 259500
+      "learning_rate": 3.696657689431787e-06,
+      "loss": 2.828,
+      "step": 279900
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.6678434130165524e-06,
-      "loss": 2.6858,
-      "step": 259600
+      "learning_rate": 3.6801148733151147e-06,
+      "loss": 2.8359,
+      "step": 280000
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.6499958950708728e-06,
-      "loss": 2.6867,
-      "step": 259700
+      "learning_rate": 3.663572057198441e-06,
+      "loss": 2.8361,
+      "step": 280100
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.6321483771251936e-06,
-      "loss": 2.688,
-      "step": 259800
+      "learning_rate": 3.647029241081768e-06,
+      "loss": 2.8409,
+      "step": 280200
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.614300859179514e-06,
-      "loss": 2.689,
-      "step": 259900
+      "learning_rate": 3.6304864249650946e-06,
+      "loss": 2.8253,
+      "step": 280300
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.5964533412338347e-06,
-      "loss": 2.6853,
-      "step": 260000
+      "learning_rate": 3.6139436088484217e-06,
+      "loss": 2.8372,
+      "step": 280400
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.578605823288155e-06,
-      "loss": 2.6822,
-      "step": 260100
+      "learning_rate": 3.5974007927317483e-06,
+      "loss": 2.8243,
+      "step": 280500
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.5607583053424763e-06,
-      "loss": 2.6965,
-      "step": 260200
+      "learning_rate": 3.5808579766150754e-06,
+      "loss": 2.8315,
+      "step": 280600
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.542910787396797e-06,
-      "loss": 2.6966,
-      "step": 260300
+      "learning_rate": 3.564315160498402e-06,
+      "loss": 2.8328,
+      "step": 280700
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.5250632694511175e-06,
-      "loss": 2.6908,
-      "step": 260400
+      "learning_rate": 3.547772344381729e-06,
+      "loss": 2.828,
+      "step": 280800
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.5072157515054382e-06,
-      "loss": 2.6828,
-      "step": 260500
+      "learning_rate": 3.531229528265056e-06,
+      "loss": 2.826,
+      "step": 280900
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.489368233559759e-06,
-      "loss": 2.6718,
-      "step": 260600
+      "learning_rate": 3.5146867121483825e-06,
+      "loss": 2.8218,
+      "step": 281000
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.4715207156140794e-06,
-      "loss": 2.686,
-      "step": 260700
+      "learning_rate": 3.4981438960317095e-06,
+      "loss": 2.8356,
+      "step": 281100
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.4536731976684006e-06,
-      "loss": 2.6885,
-      "step": 260800
+      "learning_rate": 3.481601079915036e-06,
+      "loss": 2.838,
+      "step": 281200
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.4358256797227214e-06,
-      "loss": 2.6732,
-      "step": 260900
+      "learning_rate": 3.4650582637983633e-06,
+      "loss": 2.841,
+      "step": 281300
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.4179781617770418e-06,
-      "loss": 2.68,
-      "step": 261000
+      "learning_rate": 3.44851544768169e-06,
+      "loss": 2.8254,
+      "step": 281400
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.4001306438313626e-06,
-      "loss": 2.6809,
-      "step": 261100
+      "learning_rate": 3.431972631565017e-06,
+      "loss": 2.8395,
+      "step": 281500
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.382283125885683e-06,
-      "loss": 2.6972,
-      "step": 261200
+      "learning_rate": 3.4154298154483432e-06,
+      "loss": 2.8246,
+      "step": 281600
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.3644356079400037e-06,
-      "loss": 2.6898,
-      "step": 261300
+      "learning_rate": 3.3988869993316707e-06,
+      "loss": 2.8297,
+      "step": 281700
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.346588089994325e-06,
-      "loss": 2.6904,
-      "step": 261400
+      "learning_rate": 3.382344183214997e-06,
+      "loss": 2.8368,
+      "step": 281800
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.3287405720486457e-06,
-      "loss": 2.6834,
-      "step": 261500
+      "learning_rate": 3.3658013670983244e-06,
+      "loss": 2.8315,
+      "step": 281900
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.310893054102966e-06,
-      "loss": 2.6889,
-      "step": 261600
+      "learning_rate": 3.3492585509816507e-06,
+      "loss": 2.8415,
+      "step": 282000
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.293045536157287e-06,
-      "loss": 2.6897,
-      "step": 261700
+      "learning_rate": 3.332715734864978e-06,
+      "loss": 2.8195,
+      "step": 282100
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.2751980182116073e-06,
-      "loss": 2.6902,
-      "step": 261800
+      "learning_rate": 3.3161729187483044e-06,
+      "loss": 2.8313,
+      "step": 282200
     },
     {
       "epoch": 0.93,
-      "learning_rate": 3.257350500265928e-06,
-      "loss": 2.7015,
-      "step": 261900
+      "learning_rate": 3.299630102631631e-06,
+      "loss": 2.8383,
+      "step": 282300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.283087286514958e-06,
+      "loss": 2.8421,
+      "step": 282400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2665444703982848e-06,
+      "loss": 2.8338,
+      "step": 282500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.250001654281612e-06,
+      "loss": 2.837,
+      "step": 282600
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.2395029823202484e-06,
-      "loss": 2.6892,
-      "step": 262000
+      "learning_rate": 3.2334588381649385e-06,
+      "loss": 2.8264,
+      "step": 282700
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.22165546437457e-06,
-      "loss": 2.6933,
-      "step": 262100
+      "learning_rate": 3.2169160220482656e-06,
+      "loss": 2.8345,
+      "step": 282800
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.2038079464288904e-06,
-      "loss": 2.6754,
-      "step": 262200
+      "learning_rate": 3.2003732059315922e-06,
+      "loss": 2.8433,
+      "step": 282900
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.1859604284832112e-06,
-      "loss": 2.6942,
-      "step": 262300
+      "learning_rate": 3.1838303898149193e-06,
+      "loss": 2.8395,
+      "step": 283000
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.1681129105375316e-06,
-      "loss": 2.6842,
-      "step": 262400
+      "learning_rate": 3.167287573698246e-06,
+      "loss": 2.8316,
+      "step": 283100
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.1502653925918524e-06,
-      "loss": 2.676,
-      "step": 262500
+      "learning_rate": 3.150744757581573e-06,
+      "loss": 2.8516,
+      "step": 283200
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.1324178746461727e-06,
-      "loss": 2.6829,
-      "step": 262600
+      "learning_rate": 3.1342019414648997e-06,
+      "loss": 2.8212,
+      "step": 283300
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.114570356700494e-06,
-      "loss": 2.6774,
-      "step": 262700
+      "learning_rate": 3.1176591253482263e-06,
+      "loss": 2.8262,
+      "step": 283400
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.0967228387548143e-06,
-      "loss": 2.6876,
-      "step": 262800
+      "learning_rate": 3.1011163092315534e-06,
+      "loss": 2.8288,
+      "step": 283500
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.078875320809135e-06,
-      "loss": 2.6815,
-      "step": 262900
+      "learning_rate": 3.08457349311488e-06,
+      "loss": 2.8205,
+      "step": 283600
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.061027802863456e-06,
-      "loss": 2.6745,
-      "step": 263000
+      "learning_rate": 3.068030676998207e-06,
+      "loss": 2.8384,
+      "step": 283700
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.0431802849177767e-06,
-      "loss": 2.6781,
-      "step": 263100
+      "learning_rate": 3.051487860881534e-06,
+      "loss": 2.8347,
+      "step": 283800
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.0253327669720975e-06,
-      "loss": 2.6946,
-      "step": 263200
+      "learning_rate": 3.034945044764861e-06,
+      "loss": 2.8473,
+      "step": 283900
     },
     {
       "epoch": 0.94,
-      "learning_rate": 3.007485249026418e-06,
-      "loss": 2.6873,
-      "step": 263300
+      "learning_rate": 3.018402228648187e-06,
+      "loss": 2.8355,
+      "step": 284000
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.9896377310807387e-06,
-      "loss": 2.6945,
-      "step": 263400
+      "learning_rate": 3.001859412531514e-06,
+      "loss": 2.8433,
+      "step": 284100
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.9717902131350594e-06,
-      "loss": 2.6972,
-      "step": 263500
+      "learning_rate": 2.985316596414841e-06,
+      "loss": 2.837,
+      "step": 284200
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.9539426951893802e-06,
-      "loss": 2.6925,
-      "step": 263600
+      "learning_rate": 2.968773780298168e-06,
+      "loss": 2.8173,
+      "step": 284300
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.9360951772437006e-06,
-      "loss": 2.6725,
-      "step": 263700
+      "learning_rate": 2.9522309641814946e-06,
+      "loss": 2.8312,
+      "step": 284400
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.918247659298022e-06,
-      "loss": 2.6786,
-      "step": 263800
+      "learning_rate": 2.9356881480648216e-06,
+      "loss": 2.8354,
+      "step": 284500
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.900400141352342e-06,
-      "loss": 2.6985,
-      "step": 263900
+      "learning_rate": 2.9191453319481483e-06,
+      "loss": 2.8306,
+      "step": 284600
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.882552623406663e-06,
-      "loss": 2.6911,
-      "step": 264000
+      "learning_rate": 2.9026025158314754e-06,
+      "loss": 2.8371,
+      "step": 284700
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.8647051054609833e-06,
-      "loss": 2.6892,
-      "step": 264100
+      "learning_rate": 2.886059699714802e-06,
+      "loss": 2.8415,
+      "step": 284800
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.8468575875153046e-06,
-      "loss": 2.6764,
-      "step": 264200
+      "learning_rate": 2.869516883598129e-06,
+      "loss": 2.8343,
+      "step": 284900
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.829010069569625e-06,
-      "loss": 2.6838,
-      "step": 264300
+      "learning_rate": 2.8529740674814557e-06,
+      "loss": 2.8388,
+      "step": 285000
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.8111625516239457e-06,
-      "loss": 2.6867,
-      "step": 264400
+      "learning_rate": 2.836431251364783e-06,
+      "loss": 2.8228,
+      "step": 285100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.819888435248109e-06,
+      "loss": 2.8277,
+      "step": 285200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.803345619131436e-06,
+      "loss": 2.8394,
+      "step": 285300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.7868028030147628e-06,
+      "loss": 2.8377,
+      "step": 285400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.77025998689809e-06,
+      "loss": 2.8242,
+      "step": 285500
     },
     {
       "epoch": 0.94,
-      "learning_rate": 2.7933150336782665e-06,
-      "loss": 2.6797,
-      "step": 264500
+      "learning_rate": 2.7537171707814165e-06,
+      "loss": 2.8406,
+      "step": 285600
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.7754675157325873e-06,
-      "loss": 2.6744,
-      "step": 264600
+      "epoch": 0.95,
+      "learning_rate": 2.7371743546647436e-06,
+      "loss": 2.8346,
+      "step": 285700
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 2.7576199977869077e-06,
-      "loss": 2.6841,
-      "step": 264700
+      "epoch": 0.95,
+      "learning_rate": 2.7206315385480702e-06,
+      "loss": 2.8446,
+      "step": 285800
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.739772479841229e-06,
-      "loss": 2.6936,
-      "step": 264800
+      "learning_rate": 2.7040887224313973e-06,
+      "loss": 2.8182,
+      "step": 285900
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.7219249618955493e-06,
-      "loss": 2.6896,
-      "step": 264900
+      "learning_rate": 2.687545906314724e-06,
+      "loss": 2.8417,
+      "step": 286000
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.70407744394987e-06,
-      "loss": 2.6804,
-      "step": 265000
+      "learning_rate": 2.6710030901980506e-06,
+      "loss": 2.8405,
+      "step": 286100
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.686229926004191e-06,
-      "loss": 2.6878,
-      "step": 265100
+      "learning_rate": 2.6544602740813777e-06,
+      "loss": 2.8473,
+      "step": 286200
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.6683824080585116e-06,
-      "loss": 2.6837,
-      "step": 265200
+      "learning_rate": 2.6379174579647043e-06,
+      "loss": 2.8308,
+      "step": 286300
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.650534890112832e-06,
-      "loss": 2.6931,
-      "step": 265300
+      "learning_rate": 2.621374641848031e-06,
+      "loss": 2.8271,
+      "step": 286400
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.6326873721671528e-06,
-      "loss": 2.6771,
-      "step": 265400
+      "learning_rate": 2.604831825731358e-06,
+      "loss": 2.8234,
+      "step": 286500
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.6148398542214736e-06,
-      "loss": 2.6959,
-      "step": 265500
+      "learning_rate": 2.5882890096146847e-06,
+      "loss": 2.8299,
+      "step": 286600
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.5969923362757944e-06,
-      "loss": 2.6821,
-      "step": 265600
+      "learning_rate": 2.571746193498012e-06,
+      "loss": 2.8254,
+      "step": 286700
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.5791448183301147e-06,
-      "loss": 2.6785,
-      "step": 265700
+      "learning_rate": 2.5552033773813384e-06,
+      "loss": 2.8386,
+      "step": 286800
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.5612973003844355e-06,
-      "loss": 2.6831,
-      "step": 265800
+      "learning_rate": 2.538660561264665e-06,
+      "loss": 2.8395,
+      "step": 286900
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.5434497824387563e-06,
-      "loss": 2.6949,
-      "step": 265900
+      "learning_rate": 2.522117745147992e-06,
+      "loss": 2.8335,
+      "step": 287000
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.525602264493077e-06,
-      "loss": 2.7065,
-      "step": 266000
+      "learning_rate": 2.505574929031319e-06,
+      "loss": 2.8257,
+      "step": 287100
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.507754746547398e-06,
-      "loss": 2.6873,
-      "step": 266100
+      "learning_rate": 2.489032112914646e-06,
+      "loss": 2.8402,
+      "step": 287200
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.4899072286017183e-06,
-      "loss": 2.688,
-      "step": 266200
+      "learning_rate": 2.4724892967979726e-06,
+      "loss": 2.821,
+      "step": 287300
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.472059710656039e-06,
-      "loss": 2.697,
-      "step": 266300
+      "learning_rate": 2.4559464806812996e-06,
+      "loss": 2.8185,
+      "step": 287400
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.45421219271036e-06,
-      "loss": 2.6867,
-      "step": 266400
+      "learning_rate": 2.4394036645646263e-06,
+      "loss": 2.8286,
+      "step": 287500
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.4363646747646806e-06,
-      "loss": 2.6913,
-      "step": 266500
+      "learning_rate": 2.4228608484479534e-06,
+      "loss": 2.8345,
+      "step": 287600
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.418517156819001e-06,
-      "loss": 2.6789,
-      "step": 266600
+      "learning_rate": 2.4063180323312796e-06,
+      "loss": 2.8309,
+      "step": 287700
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.4006696388733222e-06,
-      "loss": 2.6866,
-      "step": 266700
+      "learning_rate": 2.3897752162146067e-06,
+      "loss": 2.8317,
+      "step": 287800
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.3828221209276426e-06,
-      "loss": 2.6865,
-      "step": 266800
+      "learning_rate": 2.3732324000979333e-06,
+      "loss": 2.8162,
+      "step": 287900
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.3649746029819634e-06,
-      "loss": 2.6802,
-      "step": 266900
+      "learning_rate": 2.3566895839812604e-06,
+      "loss": 2.8288,
+      "step": 288000
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.347127085036284e-06,
-      "loss": 2.6881,
-      "step": 267000
+      "learning_rate": 2.340146767864587e-06,
+      "loss": 2.8258,
+      "step": 288100
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.329279567090605e-06,
-      "loss": 2.6901,
-      "step": 267100
+      "learning_rate": 2.323603951747914e-06,
+      "loss": 2.8313,
+      "step": 288200
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.3114320491449253e-06,
-      "loss": 2.6825,
-      "step": 267200
+      "learning_rate": 2.3070611356312408e-06,
+      "loss": 2.8277,
+      "step": 288300
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.293584531199246e-06,
-      "loss": 2.6923,
-      "step": 267300
+      "learning_rate": 2.290518319514568e-06,
+      "loss": 2.8345,
+      "step": 288400
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.275737013253567e-06,
-      "loss": 2.6966,
-      "step": 267400
+      "learning_rate": 2.2739755033978945e-06,
+      "loss": 2.8303,
+      "step": 288500
     },
     {
       "epoch": 0.95,
-      "learning_rate": 2.2578894953078877e-06,
-      "loss": 2.6804,
-      "step": 267500
+      "learning_rate": 2.2574326872812216e-06,
+      "loss": 2.842,
+      "step": 288600
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.240041977362208e-06,
-      "loss": 2.69,
-      "step": 267600
+      "learning_rate": 2.2408898711645482e-06,
+      "loss": 2.8173,
+      "step": 288700
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.2221944594165293e-06,
-      "loss": 2.6876,
-      "step": 267700
+      "learning_rate": 2.2243470550478753e-06,
+      "loss": 2.827,
+      "step": 288800
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.2043469414708497e-06,
-      "loss": 2.6957,
-      "step": 267800
+      "learning_rate": 2.2078042389312015e-06,
+      "loss": 2.8212,
+      "step": 288900
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.1864994235251705e-06,
-      "loss": 2.6904,
-      "step": 267900
+      "learning_rate": 2.1912614228145286e-06,
+      "loss": 2.8276,
+      "step": 289000
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.1686519055794912e-06,
-      "loss": 2.6903,
-      "step": 268000
+      "learning_rate": 2.1747186066978553e-06,
+      "loss": 2.8292,
+      "step": 289100
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.150804387633812e-06,
-      "loss": 2.6769,
-      "step": 268100
+      "learning_rate": 2.1581757905811823e-06,
+      "loss": 2.8286,
+      "step": 289200
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.1329568696881324e-06,
-      "loss": 2.6971,
-      "step": 268200
+      "learning_rate": 2.141632974464509e-06,
+      "loss": 2.8286,
+      "step": 289300
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.115109351742453e-06,
-      "loss": 2.6879,
-      "step": 268300
+      "learning_rate": 2.125090158347836e-06,
+      "loss": 2.8303,
+      "step": 289400
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.097261833796774e-06,
-      "loss": 2.68,
-      "step": 268400
+      "learning_rate": 2.1085473422311627e-06,
+      "loss": 2.8294,
+      "step": 289500
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.0794143158510948e-06,
-      "loss": 2.6877,
-      "step": 268500
+      "learning_rate": 2.09200452611449e-06,
+      "loss": 2.8348,
+      "step": 289600
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.0615667979054156e-06,
-      "loss": 2.686,
-      "step": 268600
+      "learning_rate": 2.0754617099978164e-06,
+      "loss": 2.8237,
+      "step": 289700
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.043719279959736e-06,
-      "loss": 2.6775,
-      "step": 268700
+      "learning_rate": 2.0589188938811435e-06,
+      "loss": 2.8397,
+      "step": 289800
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.0258717620140567e-06,
-      "loss": 2.6828,
-      "step": 268800
+      "learning_rate": 2.04237607776447e-06,
+      "loss": 2.8203,
+      "step": 289900
     },
     {
       "epoch": 0.96,
-      "learning_rate": 2.0080242440683775e-06,
-      "loss": 2.6877,
-      "step": 268900
+      "learning_rate": 2.0258332616477972e-06,
+      "loss": 2.8306,
+      "step": 290000
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.9901767261226983e-06,
-      "loss": 2.6853,
-      "step": 269000
+      "learning_rate": 2.0092904455311235e-06,
+      "loss": 2.8229,
+      "step": 290100
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.9723292081770187e-06,
-      "loss": 2.6961,
-      "step": 269100
+      "learning_rate": 1.9927476294144505e-06,
+      "loss": 2.8302,
+      "step": 290200
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.9544816902313395e-06,
-      "loss": 2.6897,
-      "step": 269200
+      "learning_rate": 1.976204813297777e-06,
+      "loss": 2.8368,
+      "step": 290300
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.9366341722856603e-06,
-      "loss": 2.7004,
-      "step": 269300
+      "learning_rate": 1.9596619971811043e-06,
+      "loss": 2.8335,
+      "step": 290400
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.918786654339981e-06,
-      "loss": 2.6924,
-      "step": 269400
+      "learning_rate": 1.943119181064431e-06,
+      "loss": 2.8354,
+      "step": 290500
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.9009391363943016e-06,
-      "loss": 2.6848,
-      "step": 269500
+      "learning_rate": 1.926576364947758e-06,
+      "loss": 2.8253,
+      "step": 290600
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.8830916184486226e-06,
-      "loss": 2.6926,
-      "step": 269600
+      "learning_rate": 1.9100335488310847e-06,
+      "loss": 2.8198,
+      "step": 290700
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.8652441005029432e-06,
-      "loss": 2.6867,
-      "step": 269700
+      "learning_rate": 1.8934907327144115e-06,
+      "loss": 2.82,
+      "step": 290800
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.8473965825572638e-06,
-      "loss": 2.6938,
-      "step": 269800
+      "learning_rate": 1.8769479165977384e-06,
+      "loss": 2.8286,
+      "step": 290900
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.8295490646115846e-06,
-      "loss": 2.6975,
-      "step": 269900
+      "learning_rate": 1.8604051004810652e-06,
+      "loss": 2.8221,
+      "step": 291000
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.8117015466659054e-06,
-      "loss": 2.687,
-      "step": 270000
+      "learning_rate": 1.8438622843643921e-06,
+      "loss": 2.8293,
+      "step": 291100
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.793854028720226e-06,
-      "loss": 2.6801,
-      "step": 270100
+      "learning_rate": 1.827319468247719e-06,
+      "loss": 2.8242,
+      "step": 291200
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.7760065107745468e-06,
-      "loss": 2.6872,
-      "step": 270200
+      "learning_rate": 1.8107766521310458e-06,
+      "loss": 2.8346,
+      "step": 291300
     },
     {
       "epoch": 0.96,
-      "learning_rate": 1.7581589928288673e-06,
-      "loss": 2.6981,
-      "step": 270300
+      "learning_rate": 1.7942338360143723e-06,
+      "loss": 2.8206,
+      "step": 291400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7776910198976991e-06,
+      "loss": 2.8268,
+      "step": 291500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.761148203781026e-06,
+      "loss": 2.8225,
+      "step": 291600
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.740311474883188e-06,
-      "loss": 2.6815,
-      "step": 270400
+      "learning_rate": 1.7446053876643529e-06,
+      "loss": 2.8052,
+      "step": 291700
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.722463956937509e-06,
-      "loss": 2.685,
-      "step": 270500
+      "learning_rate": 1.7280625715476797e-06,
+      "loss": 2.8306,
+      "step": 291800
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.7046164389918295e-06,
-      "loss": 2.6959,
-      "step": 270600
+      "learning_rate": 1.7115197554310066e-06,
+      "loss": 2.8316,
+      "step": 291900
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.68676892104615e-06,
-      "loss": 2.6866,
-      "step": 270700
+      "learning_rate": 1.6949769393143335e-06,
+      "loss": 2.8416,
+      "step": 292000
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.6689214031004707e-06,
-      "loss": 2.6828,
-      "step": 270800
+      "learning_rate": 1.6784341231976603e-06,
+      "loss": 2.8366,
+      "step": 292100
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.6510738851547917e-06,
-      "loss": 2.6883,
-      "step": 270900
+      "learning_rate": 1.6618913070809872e-06,
+      "loss": 2.8356,
+      "step": 292200
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.6332263672091122e-06,
-      "loss": 2.6851,
-      "step": 271000
+      "learning_rate": 1.645348490964314e-06,
+      "loss": 2.849,
+      "step": 292300
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.6153788492634328e-06,
-      "loss": 2.6877,
-      "step": 271100
+      "learning_rate": 1.628805674847641e-06,
+      "loss": 2.8333,
+      "step": 292400
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.5975313313177538e-06,
-      "loss": 2.6836,
-      "step": 271200
+      "learning_rate": 1.6122628587309678e-06,
+      "loss": 2.8314,
+      "step": 292500
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.5796838133720744e-06,
-      "loss": 2.6864,
-      "step": 271300
+      "learning_rate": 1.5957200426142942e-06,
+      "loss": 2.8259,
+      "step": 292600
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.5618362954263952e-06,
-      "loss": 2.68,
-      "step": 271400
+      "learning_rate": 1.579177226497621e-06,
+      "loss": 2.8291,
+      "step": 292700
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.5439887774807158e-06,
-      "loss": 2.6907,
-      "step": 271500
+      "learning_rate": 1.562634410380948e-06,
+      "loss": 2.833,
+      "step": 292800
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.5261412595350366e-06,
-      "loss": 2.6813,
-      "step": 271600
+      "learning_rate": 1.5460915942642748e-06,
+      "loss": 2.811,
+      "step": 292900
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.5082937415893574e-06,
-      "loss": 2.682,
-      "step": 271700
+      "learning_rate": 1.5295487781476017e-06,
+      "loss": 2.8259,
+      "step": 293000
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.490446223643678e-06,
-      "loss": 2.6878,
-      "step": 271800
+      "learning_rate": 1.5130059620309285e-06,
+      "loss": 2.8226,
+      "step": 293100
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.4725987056979987e-06,
-      "loss": 2.6864,
-      "step": 271900
+      "learning_rate": 1.4964631459142554e-06,
+      "loss": 2.8374,
+      "step": 293200
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.4547511877523193e-06,
-      "loss": 2.6798,
-      "step": 272000
+      "learning_rate": 1.4799203297975823e-06,
+      "loss": 2.8259,
+      "step": 293300
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.43690366980664e-06,
-      "loss": 2.6884,
-      "step": 272100
+      "learning_rate": 1.4633775136809091e-06,
+      "loss": 2.8263,
+      "step": 293400
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.4190561518609609e-06,
-      "loss": 2.6844,
-      "step": 272200
+      "learning_rate": 1.4468346975642358e-06,
+      "loss": 2.8241,
+      "step": 293500
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.4012086339152815e-06,
-      "loss": 2.6828,
-      "step": 272300
+      "learning_rate": 1.4302918814475627e-06,
+      "loss": 2.8238,
+      "step": 293600
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.3833611159696023e-06,
-      "loss": 2.6909,
-      "step": 272400
+      "learning_rate": 1.4137490653308895e-06,
+      "loss": 2.8337,
+      "step": 293700
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.365513598023923e-06,
-      "loss": 2.6864,
-      "step": 272500
+      "learning_rate": 1.3972062492142164e-06,
+      "loss": 2.8257,
+      "step": 293800
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.3476660800782436e-06,
-      "loss": 2.6834,
-      "step": 272600
+      "learning_rate": 1.3806634330975432e-06,
+      "loss": 2.8205,
+      "step": 293900
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.3298185621325644e-06,
-      "loss": 2.6892,
-      "step": 272700
+      "learning_rate": 1.3641206169808701e-06,
+      "loss": 2.837,
+      "step": 294000
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.311971044186885e-06,
-      "loss": 2.6802,
-      "step": 272800
+      "learning_rate": 1.3475778008641968e-06,
+      "loss": 2.8137,
+      "step": 294100
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.2941235262412056e-06,
-      "loss": 2.684,
-      "step": 272900
+      "learning_rate": 1.3310349847475236e-06,
+      "loss": 2.8323,
+      "step": 294200
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.2762760082955264e-06,
-      "loss": 2.6775,
-      "step": 273000
+      "learning_rate": 1.3144921686308505e-06,
+      "loss": 2.8351,
+      "step": 294300
     },
     {
       "epoch": 0.97,
-      "learning_rate": 1.258428490349847e-06,
-      "loss": 2.6922,
-      "step": 273100
+      "learning_rate": 1.2979493525141774e-06,
+      "loss": 2.8194,
+      "step": 294400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2814065363975042e-06,
+      "loss": 2.8346,
+      "step": 294500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2648637202808309e-06,
+      "loss": 2.8111,
+      "step": 294600
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.2405809724041677e-06,
-      "loss": 2.6803,
-      "step": 273200
+      "learning_rate": 1.2483209041641577e-06,
+      "loss": 2.8368,
+      "step": 294700
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.2227334544584885e-06,
-      "loss": 2.6858,
-      "step": 273300
+      "learning_rate": 1.2317780880474846e-06,
+      "loss": 2.8239,
+      "step": 294800
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.2048859365128091e-06,
-      "loss": 2.6886,
-      "step": 273400
+      "learning_rate": 1.2152352719308115e-06,
+      "loss": 2.8217,
+      "step": 294900
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.18703841856713e-06,
-      "loss": 2.6808,
-      "step": 273500
+      "learning_rate": 1.1986924558141381e-06,
+      "loss": 2.8334,
+      "step": 295000
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.1691909006214505e-06,
-      "loss": 2.6876,
-      "step": 273600
+      "learning_rate": 1.182149639697465e-06,
+      "loss": 2.8206,
+      "step": 295100
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.1513433826757713e-06,
-      "loss": 2.6869,
-      "step": 273700
+      "learning_rate": 1.1656068235807918e-06,
+      "loss": 2.8312,
+      "step": 295200
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.133495864730092e-06,
-      "loss": 2.6812,
-      "step": 273800
+      "learning_rate": 1.1490640074641187e-06,
+      "loss": 2.8306,
+      "step": 295300
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.1156483467844126e-06,
-      "loss": 2.6768,
-      "step": 273900
+      "learning_rate": 1.1325211913474454e-06,
+      "loss": 2.8268,
+      "step": 295400
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.0978008288387334e-06,
-      "loss": 2.6939,
-      "step": 274000
+      "learning_rate": 1.1159783752307722e-06,
+      "loss": 2.828,
+      "step": 295500
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.0799533108930542e-06,
-      "loss": 2.6957,
-      "step": 274100
+      "learning_rate": 1.099435559114099e-06,
+      "loss": 2.8254,
+      "step": 295600
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.0621057929473748e-06,
-      "loss": 2.6826,
-      "step": 274200
+      "learning_rate": 1.082892742997426e-06,
+      "loss": 2.8146,
+      "step": 295700
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.0442582750016956e-06,
-      "loss": 2.681,
-      "step": 274300
+      "learning_rate": 1.0663499268807528e-06,
+      "loss": 2.826,
+      "step": 295800
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.0264107570560162e-06,
-      "loss": 2.6763,
-      "step": 274400
+      "learning_rate": 1.0498071107640797e-06,
+      "loss": 2.8366,
+      "step": 295900
     },
     {
       "epoch": 0.98,
-      "learning_rate": 1.008563239110337e-06,
-      "loss": 2.6828,
-      "step": 274500
+      "learning_rate": 1.0332642946474063e-06,
+      "loss": 2.8163,
+      "step": 296000
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.907157211646578e-07,
-      "loss": 2.6802,
-      "step": 274600
+      "learning_rate": 1.0167214785307332e-06,
+      "loss": 2.8315,
+      "step": 296100
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.728682032189783e-07,
-      "loss": 2.6733,
-      "step": 274700
+      "learning_rate": 1.00017866241406e-06,
+      "loss": 2.8404,
+      "step": 296200
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.550206852732991e-07,
-      "loss": 2.6931,
-      "step": 274800
+      "learning_rate": 9.83635846297387e-07,
+      "loss": 2.8324,
+      "step": 296300
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.371731673276198e-07,
-      "loss": 2.6864,
-      "step": 274900
+      "learning_rate": 9.670930301807138e-07,
+      "loss": 2.8156,
+      "step": 296400
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.193256493819405e-07,
-      "loss": 2.6891,
-      "step": 275000
+      "learning_rate": 9.505502140640406e-07,
+      "loss": 2.8378,
+      "step": 296500
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.014781314362612e-07,
-      "loss": 2.686,
-      "step": 275100
+      "learning_rate": 9.340073979473674e-07,
+      "loss": 2.8393,
+      "step": 296600
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.836306134905818e-07,
-      "loss": 2.6796,
-      "step": 275200
+      "learning_rate": 9.174645818306942e-07,
+      "loss": 2.8294,
+      "step": 296700
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.657830955449026e-07,
-      "loss": 2.6936,
-      "step": 275300
+      "learning_rate": 9.00921765714021e-07,
+      "loss": 2.8253,
+      "step": 296800
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.479355775992234e-07,
-      "loss": 2.6868,
-      "step": 275400
+      "learning_rate": 8.843789495973479e-07,
+      "loss": 2.8504,
+      "step": 296900
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.300880596535439e-07,
-      "loss": 2.6777,
-      "step": 275500
+      "learning_rate": 8.678361334806748e-07,
+      "loss": 2.8108,
+      "step": 297000
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.122405417078647e-07,
-      "loss": 2.6859,
-      "step": 275600
+      "learning_rate": 8.512933173640016e-07,
+      "loss": 2.8242,
+      "step": 297100
     },
     {
       "epoch": 0.98,
-      "learning_rate": 7.943930237621855e-07,
-      "loss": 2.6915,
-      "step": 275700
+      "learning_rate": 8.347505012473283e-07,
+      "loss": 2.828,
+      "step": 297200
     },
     {
       "epoch": 0.98,
-      "learning_rate": 7.765455058165062e-07,
-      "loss": 2.6764,
-      "step": 275800
+      "learning_rate": 8.182076851306551e-07,
+      "loss": 2.8393,
+      "step": 297300
     },
     {
       "epoch": 0.98,
-      "learning_rate": 7.586979878708269e-07,
-      "loss": 2.6742,
-      "step": 275900
+      "learning_rate": 8.01664869013982e-07,
+      "loss": 2.8163,
+      "step": 297400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.851220528973089e-07,
+      "loss": 2.8325,
+      "step": 297500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.685792367806356e-07,
+      "loss": 2.835,
+      "step": 297600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.520364206639625e-07,
+      "loss": 2.8243,
+      "step": 297700
     },
     {
       "epoch": 0.99,
-      "learning_rate": 7.408504699251476e-07,
-      "loss": 2.6871,
-      "step": 276000
+      "learning_rate": 7.354936045472893e-07,
+      "loss": 2.8446,
+      "step": 297800
     },
     {
       "epoch": 0.99,
-      "learning_rate": 7.230029519794683e-07,
-      "loss": 2.6726,
-      "step": 276100
+      "learning_rate": 7.189507884306161e-07,
+      "loss": 2.8212,
+      "step": 297900
     },
     {
       "epoch": 0.99,
-      "learning_rate": 7.05155434033789e-07,
-      "loss": 2.6765,
-      "step": 276200
+      "learning_rate": 7.02407972313943e-07,
+      "loss": 2.8442,
+      "step": 298000
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.873079160881096e-07,
-      "loss": 2.6749,
-      "step": 276300
+      "learning_rate": 6.858651561972698e-07,
+      "loss": 2.8323,
+      "step": 298100
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.694603981424303e-07,
-      "loss": 2.685,
-      "step": 276400
+      "learning_rate": 6.693223400805967e-07,
+      "loss": 2.823,
+      "step": 298200
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.51612880196751e-07,
-      "loss": 2.6819,
-      "step": 276500
+      "learning_rate": 6.527795239639235e-07,
+      "loss": 2.8213,
+      "step": 298300
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.337653622510718e-07,
-      "loss": 2.6861,
-      "step": 276600
+      "learning_rate": 6.362367078472503e-07,
+      "loss": 2.835,
+      "step": 298400
     },
     {
       "epoch": 0.99,
-      "learning_rate": 6.159178443053925e-07,
-      "loss": 2.6946,
-      "step": 276700
+      "learning_rate": 6.196938917305772e-07,
+      "loss": 2.8275,
+      "step": 298500
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.980703263597132e-07,
-      "loss": 2.6903,
-      "step": 276800
+      "learning_rate": 6.031510756139039e-07,
+      "loss": 2.8307,
+      "step": 298600
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.802228084140338e-07,
-      "loss": 2.6886,
-      "step": 276900
+      "learning_rate": 5.866082594972308e-07,
+      "loss": 2.8284,
+      "step": 298700
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.623752904683546e-07,
-      "loss": 2.6833,
-      "step": 277000
+      "learning_rate": 5.700654433805577e-07,
+      "loss": 2.8197,
+      "step": 298800
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.445277725226753e-07,
-      "loss": 2.6829,
-      "step": 277100
+      "learning_rate": 5.535226272638844e-07,
+      "loss": 2.8304,
+      "step": 298900
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.26680254576996e-07,
-      "loss": 2.69,
-      "step": 277200
+      "learning_rate": 5.369798111472113e-07,
+      "loss": 2.8341,
+      "step": 299000
     },
     {
       "epoch": 0.99,
-      "learning_rate": 5.088327366313167e-07,
-      "loss": 2.6866,
-      "step": 277300
+      "learning_rate": 5.20436995030538e-07,
+      "loss": 2.8348,
+      "step": 299100
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.909852186856375e-07,
-      "loss": 2.6755,
-      "step": 277400
+      "learning_rate": 5.038941789138649e-07,
+      "loss": 2.8334,
+      "step": 299200
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.731377007399581e-07,
-      "loss": 2.6973,
-      "step": 277500
+      "learning_rate": 4.873513627971917e-07,
+      "loss": 2.8271,
+      "step": 299300
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.552901827942788e-07,
-      "loss": 2.6872,
-      "step": 277600
+      "learning_rate": 4.708085466805186e-07,
+      "loss": 2.8332,
+      "step": 299400
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.374426648485995e-07,
-      "loss": 2.6886,
-      "step": 277700
+      "learning_rate": 4.5426573056384535e-07,
+      "loss": 2.837,
+      "step": 299500
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.195951469029203e-07,
-      "loss": 2.697,
-      "step": 277800
+      "learning_rate": 4.377229144471722e-07,
+      "loss": 2.8345,
+      "step": 299600
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.0174762895724097e-07,
-      "loss": 2.6859,
-      "step": 277900
+      "learning_rate": 4.211800983304991e-07,
+      "loss": 2.8161,
+      "step": 299700
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.8390011101156165e-07,
-      "loss": 2.7017,
-      "step": 278000
+      "learning_rate": 4.0463728221382583e-07,
+      "loss": 2.8552,
+      "step": 299800
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.6605259306588234e-07,
-      "loss": 2.6748,
-      "step": 278100
+      "learning_rate": 3.880944660971527e-07,
+      "loss": 2.8121,
+      "step": 299900
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.48205075120203e-07,
-      "loss": 2.6919,
-      "step": 278200
+      "learning_rate": 3.715516499804795e-07,
+      "loss": 2.8275,
+      "step": 300000
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.3035755717452376e-07,
-      "loss": 2.6853,
-      "step": 278300
+      "learning_rate": 3.550088338638063e-07,
+      "loss": 2.8354,
+      "step": 300100
     },
     {
       "epoch": 0.99,
-      "learning_rate": 3.1251003922884445e-07,
-      "loss": 2.6843,
-      "step": 278400
+      "learning_rate": 3.3846601774713313e-07,
+      "loss": 2.8394,
+      "step": 300200
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.946625212831652e-07,
-      "loss": 2.6747,
-      "step": 278500
+      "learning_rate": 3.2192320163045994e-07,
+      "loss": 2.8358,
+      "step": 300300
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.7681500333748587e-07,
-      "loss": 2.6838,
-      "step": 278600
+      "learning_rate": 3.053803855137868e-07,
+      "loss": 2.8343,
+      "step": 300400
     },
     {
       "epoch": 0.99,
-      "learning_rate": 2.589674853918066e-07,
-      "loss": 2.6912,
-      "step": 278700
+      "learning_rate": 2.888375693971136e-07,
+      "loss": 2.8372,
+      "step": 300500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7229475328044043e-07,
+      "loss": 2.8247,
+      "step": 300600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.557519371637673e-07,
+      "loss": 2.8358,
+      "step": 300700
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.411199674461273e-07,
-      "loss": 2.6809,
-      "step": 278800
+      "learning_rate": 2.392091210470941e-07,
+      "loss": 2.8383,
+      "step": 300800
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.23272449500448e-07,
-      "loss": 2.6763,
-      "step": 278900
+      "learning_rate": 2.2266630493042091e-07,
+      "loss": 2.8191,
+      "step": 300900
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.0542493155476867e-07,
-      "loss": 2.6928,
-      "step": 279000
+      "learning_rate": 2.0612348881374775e-07,
+      "loss": 2.8207,
+      "step": 301000
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.8757741360908938e-07,
-      "loss": 2.6839,
-      "step": 279100
+      "learning_rate": 1.8958067269707456e-07,
+      "loss": 2.8157,
+      "step": 301100
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.697298956634101e-07,
-      "loss": 2.6872,
-      "step": 279200
+      "learning_rate": 1.730378565804014e-07,
+      "loss": 2.8217,
+      "step": 301200
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.518823777177308e-07,
-      "loss": 2.6973,
-      "step": 279300
+      "learning_rate": 1.5649504046372824e-07,
+      "loss": 2.8379,
+      "step": 301300
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.340348597720515e-07,
-      "loss": 2.706,
-      "step": 279400
+      "learning_rate": 1.3995222434705505e-07,
+      "loss": 2.8315,
+      "step": 301400
     },
     {
       "epoch": 1.0,
-      "learning_rate": 1.1618734182637223e-07,
-      "loss": 2.6818,
-      "step": 279500
+      "learning_rate": 1.2340940823038189e-07,
+      "loss": 2.8273,
+      "step": 301500
     },
     {
       "epoch": 1.0,
-      "learning_rate": 9.833982388069292e-08,
-      "loss": 2.6777,
-      "step": 279600
+      "learning_rate": 1.0686659211370871e-07,
+      "loss": 2.8192,
+      "step": 301600
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.049230593501362e-08,
-      "loss": 2.692,
-      "step": 279700
+      "learning_rate": 9.032377599703553e-08,
+      "loss": 2.8332,
+      "step": 301700
     },
     {
       "epoch": 1.0,
-      "learning_rate": 6.264478798933432e-08,
-      "loss": 2.6734,
-      "step": 279800
+      "learning_rate": 7.378095988036236e-08,
+      "loss": 2.8372,
+      "step": 301800
     },
     {
       "epoch": 1.0,
-      "learning_rate": 4.4797270043655027e-08,
-      "loss": 2.6892,
-      "step": 279900
+      "learning_rate": 5.723814376368919e-08,
+      "loss": 2.8335,
+      "step": 301900
     },
     {
       "epoch": 1.0,
-      "learning_rate": 2.6949752097975735e-08,
-      "loss": 2.6823,
-      "step": 280000
+      "learning_rate": 4.0695327647016006e-08,
+      "loss": 2.8349,
+      "step": 302000
     }
   ],
-  "max_steps": 280151,
+  "max_steps": 302246,
   "num_train_epochs": 1,
-  "total_flos": 7.58820243932971e+18,
+  "total_flos": 8.184418510307328e+18,
   "trial_name": null,
   "trial_params": null
 }