diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,3736 +1,4096 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 6.73398680387738,
-  "global_step": 62000,
+  "epoch": 7.385674332726927,
+  "global_step": 68000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.01,
-      "learning_rate": 9.989138698816118e-05,
-      "loss": 9.1158,
+      "learning_rate": 9.989138698816118e-06,
+      "loss": 2.6315,
       "step": 100
     },
     {
       "epoch": 0.02,
-      "learning_rate": 9.978277397632236e-05,
-      "loss": 5.8307,
+      "learning_rate": 9.978277397632237e-06,
+      "loss": 2.6987,
       "step": 200
     },
     {
       "epoch": 0.03,
-      "learning_rate": 9.967416096448356e-05,
-      "loss": 5.4251,
+      "learning_rate": 9.967416096448356e-06,
+      "loss": 2.7695,
       "step": 300
     },
     {
       "epoch": 0.04,
-      "learning_rate": 9.956554795264474e-05,
-      "loss": 5.2931,
+      "learning_rate": 9.956554795264473e-06,
+      "loss": 2.7931,
       "step": 400
     },
     {
       "epoch": 0.05,
-      "learning_rate": 9.945693494080592e-05,
-      "loss": 5.0326,
+      "learning_rate": 9.94569349408059e-06,
+      "loss": 2.6996,
       "step": 500
     },
     {
       "epoch": 0.07,
-      "learning_rate": 9.93483219289671e-05,
-      "loss": 4.8397,
+      "learning_rate": 9.93483219289671e-06,
+      "loss": 2.6383,
       "step": 600
     },
     {
       "epoch": 0.08,
-      "learning_rate": 9.923970891712828e-05,
-      "loss": 4.9379,
+      "learning_rate": 9.923970891712827e-06,
+      "loss": 2.8273,
       "step": 700
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.913109590528946e-05,
-      "loss": 4.6978,
+      "learning_rate": 9.913109590528946e-06,
+      "loss": 2.6572,
       "step": 800
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.902248289345064e-05,
-      "loss": 4.7727,
+      "learning_rate": 9.902248289345065e-06,
+      "loss": 2.7393,
       "step": 900
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.891386988161182e-05,
-      "loss": 4.5856,
+      "learning_rate": 9.891386988161182e-06,
+      "loss": 2.6609,
       "step": 1000
     },
     {
       "epoch": 0.12,
-      "learning_rate": 9.8805256869773e-05,
-      "loss": 4.5636,
+      "learning_rate": 9.8805256869773e-06,
+      "loss": 2.7242,
       "step": 1100
     },
     {
       "epoch": 0.13,
-      "learning_rate": 9.869664385793419e-05,
-      "loss": 4.4165,
+      "learning_rate": 9.869664385793419e-06,
+      "loss": 2.7032,
       "step": 1200
     },
     {
       "epoch": 0.14,
-      "learning_rate": 9.858803084609537e-05,
-      "loss": 4.4187,
+      "learning_rate": 9.858803084609538e-06,
+      "loss": 2.7095,
       "step": 1300
     },
     {
       "epoch": 0.15,
-      "learning_rate": 9.847941783425655e-05,
-      "loss": 4.3553,
+      "learning_rate": 9.847941783425655e-06,
+      "loss": 2.6771,
       "step": 1400
     },
     {
       "epoch": 0.16,
-      "learning_rate": 9.837080482241773e-05,
-      "loss": 4.3312,
+      "learning_rate": 9.837080482241772e-06,
+      "loss": 2.6621,
       "step": 1500
     },
     {
       "epoch": 0.17,
-      "learning_rate": 9.826219181057891e-05,
-      "loss": 4.3841,
+      "learning_rate": 9.826219181057891e-06,
+      "loss": 2.769,
       "step": 1600
     },
     {
       "epoch": 0.18,
-      "learning_rate": 9.81535787987401e-05,
-      "loss": 4.3428,
+      "learning_rate": 9.81535787987401e-06,
+      "loss": 2.7221,
       "step": 1700
     },
     {
       "epoch": 0.2,
-      "learning_rate": 9.804496578690128e-05,
-      "loss": 4.3044,
+      "learning_rate": 9.804496578690128e-06,
+      "loss": 2.7357,
       "step": 1800
     },
     {
       "epoch": 0.21,
-      "learning_rate": 9.793635277506245e-05,
-      "loss": 4.2581,
+      "learning_rate": 9.793635277506247e-06,
+      "loss": 2.7397,
       "step": 1900
     },
     {
       "epoch": 0.22,
-      "learning_rate": 9.782773976322363e-05,
-      "loss": 4.1639,
+      "learning_rate": 9.782773976322364e-06,
+      "loss": 2.6575,
       "step": 2000
     },
     {
       "epoch": 0.23,
-      "learning_rate": 9.771912675138481e-05,
-      "loss": 4.231,
+      "learning_rate": 9.771912675138481e-06,
+      "loss": 2.7414,
       "step": 2100
     },
     {
       "epoch": 0.24,
-      "learning_rate": 9.7610513739546e-05,
-      "loss": 4.1087,
+      "learning_rate": 9.7610513739546e-06,
+      "loss": 2.6627,
       "step": 2200
     },
     {
       "epoch": 0.25,
-      "learning_rate": 9.750190072770718e-05,
-      "loss": 4.2166,
+      "learning_rate": 9.75019007277072e-06,
+      "loss": 2.7529,
       "step": 2300
     },
     {
       "epoch": 0.26,
-      "learning_rate": 9.739328771586836e-05,
-      "loss": 4.0581,
+      "learning_rate": 9.739328771586837e-06,
+      "loss": 2.6718,
       "step": 2400
     },
     {
       "epoch": 0.27,
-      "learning_rate": 9.728467470402954e-05,
-      "loss": 4.0475,
+      "learning_rate": 9.728467470402954e-06,
+      "loss": 2.6459,
       "step": 2500
     },
     {
       "epoch": 0.28,
-      "learning_rate": 9.717606169219074e-05,
-      "loss": 4.0477,
+      "learning_rate": 9.717606169219073e-06,
+      "loss": 2.6565,
       "step": 2600
     },
     {
       "epoch": 0.29,
-      "learning_rate": 9.706744868035192e-05,
-      "loss": 4.2303,
+      "learning_rate": 9.706744868035192e-06,
+      "loss": 2.8173,
       "step": 2700
     },
     {
       "epoch": 0.3,
-      "learning_rate": 9.69588356685131e-05,
-      "loss": 4.0649,
+      "learning_rate": 9.69588356685131e-06,
+      "loss": 2.7082,
       "step": 2800
     },
     {
       "epoch": 0.31,
-      "learning_rate": 9.685022265667428e-05,
-      "loss": 4.0515,
+      "learning_rate": 9.685022265667428e-06,
+      "loss": 2.7028,
       "step": 2900
     },
     {
       "epoch": 0.33,
-      "learning_rate": 9.674160964483546e-05,
-      "loss": 4.0151,
+      "learning_rate": 9.674160964483546e-06,
+      "loss": 2.6673,
       "step": 3000
     },
     {
       "epoch": 0.34,
-      "learning_rate": 9.663299663299664e-05,
-      "loss": 3.9701,
+      "learning_rate": 9.663299663299665e-06,
+      "loss": 2.6386,
       "step": 3100
     },
     {
       "epoch": 0.35,
-      "learning_rate": 9.652438362115782e-05,
-      "loss": 3.8968,
+      "learning_rate": 9.652438362115782e-06,
+      "loss": 2.6181,
       "step": 3200
     },
     {
       "epoch": 0.36,
-      "learning_rate": 9.6415770609319e-05,
-      "loss": 4.0167,
+      "learning_rate": 9.641577060931901e-06,
+      "loss": 2.7044,
       "step": 3300
     },
     {
       "epoch": 0.37,
-      "learning_rate": 9.630715759748018e-05,
-      "loss": 4.0087,
+      "learning_rate": 9.630715759748018e-06,
+      "loss": 2.7406,
       "step": 3400
     },
     {
       "epoch": 0.38,
-      "learning_rate": 9.619854458564136e-05,
-      "loss": 4.0071,
+      "learning_rate": 9.619854458564136e-06,
+      "loss": 2.7391,
       "step": 3500
     },
     {
       "epoch": 0.39,
-      "learning_rate": 9.608993157380255e-05,
-      "loss": 4.0202,
+      "learning_rate": 9.608993157380255e-06,
+      "loss": 2.7654,
       "step": 3600
     },
     {
       "epoch": 0.4,
-      "learning_rate": 9.598131856196373e-05,
-      "loss": 4.069,
+      "learning_rate": 9.598131856196374e-06,
+      "loss": 2.8028,
       "step": 3700
     },
     {
       "epoch": 0.41,
-      "learning_rate": 9.587270555012491e-05,
-      "loss": 3.9612,
+      "learning_rate": 9.587270555012491e-06,
+      "loss": 2.7452,
       "step": 3800
     },
     {
       "epoch": 0.42,
-      "learning_rate": 9.576409253828609e-05,
-      "loss": 3.8818,
+      "learning_rate": 9.57640925382861e-06,
+      "loss": 2.6812,
       "step": 3900
     },
     {
       "epoch": 0.43,
-      "learning_rate": 9.565547952644728e-05,
-      "loss": 3.9479,
+      "learning_rate": 9.565547952644729e-06,
+      "loss": 2.7176,
       "step": 4000
     },
     {
       "epoch": 0.45,
-      "learning_rate": 9.554686651460846e-05,
-      "loss": 3.8681,
+      "learning_rate": 9.554686651460846e-06,
+      "loss": 2.6697,
       "step": 4100
     },
     {
       "epoch": 0.46,
-      "learning_rate": 9.543825350276964e-05,
-      "loss": 3.8267,
+      "learning_rate": 9.543825350276964e-06,
+      "loss": 2.6532,
       "step": 4200
     },
     {
       "epoch": 0.47,
-      "learning_rate": 9.532964049093082e-05,
-      "loss": 3.9206,
+      "learning_rate": 9.532964049093083e-06,
+      "loss": 2.7343,
       "step": 4300
     },
     {
       "epoch": 0.48,
-      "learning_rate": 9.5221027479092e-05,
-      "loss": 3.8918,
+      "learning_rate": 9.5221027479092e-06,
+      "loss": 2.7422,
       "step": 4400
     },
     {
       "epoch": 0.49,
-      "learning_rate": 9.511241446725318e-05,
-      "loss": 3.7453,
+      "learning_rate": 9.511241446725319e-06,
+      "loss": 2.6157,
       "step": 4500
     },
     {
       "epoch": 0.5,
-      "learning_rate": 9.500380145541436e-05,
-      "loss": 3.8368,
+      "learning_rate": 9.500380145541436e-06,
+      "loss": 2.7115,
       "step": 4600
     },
     {
       "epoch": 0.51,
-      "learning_rate": 9.489518844357554e-05,
-      "loss": 3.8248,
+      "learning_rate": 9.489518844357555e-06,
+      "loss": 2.6865,
       "step": 4700
     },
     {
       "epoch": 0.52,
-      "learning_rate": 9.478657543173672e-05,
-      "loss": 3.8242,
+      "learning_rate": 9.478657543173672e-06,
+      "loss": 2.6929,
       "step": 4800
     },
     {
       "epoch": 0.53,
-      "learning_rate": 9.46779624198979e-05,
-      "loss": 3.8506,
+      "learning_rate": 9.46779624198979e-06,
+      "loss": 2.721,
       "step": 4900
     },
     {
       "epoch": 0.54,
-      "learning_rate": 9.45693494080591e-05,
-      "loss": 3.878,
+      "learning_rate": 9.45693494080591e-06,
+      "loss": 2.7737,
       "step": 5000
     },
     {
       "epoch": 0.55,
-      "learning_rate": 9.446073639622027e-05,
-      "loss": 3.7932,
+      "learning_rate": 9.446073639622028e-06,
+      "loss": 2.6683,
       "step": 5100
     },
     {
       "epoch": 0.56,
-      "learning_rate": 9.435212338438145e-05,
-      "loss": 3.8755,
+      "learning_rate": 9.435212338438145e-06,
+      "loss": 2.7539,
       "step": 5200
     },
     {
       "epoch": 0.58,
-      "learning_rate": 9.424351037254263e-05,
-      "loss": 3.7838,
+      "learning_rate": 9.424351037254264e-06,
+      "loss": 2.7268,
       "step": 5300
     },
     {
       "epoch": 0.59,
-      "learning_rate": 9.413489736070383e-05,
-      "loss": 3.856,
+      "learning_rate": 9.413489736070383e-06,
+      "loss": 2.7706,
       "step": 5400
     },
     {
       "epoch": 0.6,
-      "learning_rate": 9.402628434886501e-05,
-      "loss": 3.6716,
+      "learning_rate": 9.4026284348865e-06,
+      "loss": 2.6173,
       "step": 5500
     },
     {
       "epoch": 0.61,
-      "learning_rate": 9.391767133702617e-05,
-      "loss": 3.7467,
+      "learning_rate": 9.391767133702618e-06,
+      "loss": 2.6978,
       "step": 5600
     },
     {
       "epoch": 0.62,
-      "learning_rate": 9.380905832518735e-05,
-      "loss": 3.6716,
+      "learning_rate": 9.380905832518737e-06,
+      "loss": 2.6585,
       "step": 5700
     },
     {
       "epoch": 0.63,
-      "learning_rate": 9.370044531334853e-05,
-      "loss": 3.6479,
+      "learning_rate": 9.370044531334854e-06,
+      "loss": 2.6317,
       "step": 5800
     },
     {
       "epoch": 0.64,
-      "learning_rate": 9.359183230150973e-05,
-      "loss": 3.6937,
+      "learning_rate": 9.359183230150973e-06,
+      "loss": 2.6965,
       "step": 5900
     },
     {
       "epoch": 0.65,
-      "learning_rate": 9.348321928967091e-05,
-      "loss": 3.7943,
+      "learning_rate": 9.34832192896709e-06,
+      "loss": 2.79,
       "step": 6000
     },
     {
       "epoch": 0.66,
-      "learning_rate": 9.337460627783209e-05,
-      "loss": 3.6678,
+      "learning_rate": 9.33746062778321e-06,
+      "loss": 2.6627,
       "step": 6100
     },
     {
       "epoch": 0.67,
-      "learning_rate": 9.326599326599327e-05,
-      "loss": 3.6824,
+      "learning_rate": 9.326599326599327e-06,
+      "loss": 2.6838,
       "step": 6200
     },
     {
       "epoch": 0.68,
-      "learning_rate": 9.315738025415445e-05,
-      "loss": 3.6794,
+      "learning_rate": 9.315738025415446e-06,
+      "loss": 2.7199,
       "step": 6300
     },
     {
       "epoch": 0.7,
-      "learning_rate": 9.304876724231564e-05,
-      "loss": 3.7675,
+      "learning_rate": 9.304876724231565e-06,
+      "loss": 2.7603,
       "step": 6400
     },
     {
       "epoch": 0.71,
-      "learning_rate": 9.294015423047682e-05,
-      "loss": 3.6341,
+      "learning_rate": 9.294015423047682e-06,
+      "loss": 2.6606,
       "step": 6500
     },
     {
       "epoch": 0.72,
-      "learning_rate": 9.2831541218638e-05,
-      "loss": 3.6805,
+      "learning_rate": 9.2831541218638e-06,
+      "loss": 2.7007,
       "step": 6600
     },
     {
       "epoch": 0.73,
-      "learning_rate": 9.272292820679918e-05,
-      "loss": 3.6479,
+      "learning_rate": 9.272292820679918e-06,
+      "loss": 2.7015,
       "step": 6700
     },
     {
       "epoch": 0.74,
-      "learning_rate": 9.261431519496036e-05,
-      "loss": 3.6145,
+      "learning_rate": 9.261431519496037e-06,
+      "loss": 2.6503,
       "step": 6800
     },
     {
       "epoch": 0.75,
-      "learning_rate": 9.250570218312154e-05,
-      "loss": 3.5535,
+      "learning_rate": 9.250570218312155e-06,
+      "loss": 2.6384,
       "step": 6900
     },
     {
       "epoch": 0.76,
-      "learning_rate": 9.239708917128272e-05,
-      "loss": 3.6772,
+      "learning_rate": 9.239708917128272e-06,
+      "loss": 2.7407,
       "step": 7000
     },
     {
       "epoch": 0.77,
-      "learning_rate": 9.22884761594439e-05,
-      "loss": 3.6542,
+      "learning_rate": 9.228847615944391e-06,
+      "loss": 2.7046,
       "step": 7100
     },
     {
       "epoch": 0.78,
-      "learning_rate": 9.217986314760508e-05,
-      "loss": 3.5783,
+      "learning_rate": 9.217986314760508e-06,
+      "loss": 2.6495,
       "step": 7200
     },
     {
       "epoch": 0.79,
-      "learning_rate": 9.207125013576627e-05,
-      "loss": 3.5812,
+      "learning_rate": 9.207125013576627e-06,
+      "loss": 2.6634,
       "step": 7300
     },
     {
       "epoch": 0.8,
-      "learning_rate": 9.196263712392745e-05,
-      "loss": 3.5677,
+      "learning_rate": 9.196263712392746e-06,
+      "loss": 2.6955,
       "step": 7400
     },
     {
       "epoch": 0.81,
-      "learning_rate": 9.185402411208863e-05,
-      "loss": 3.5258,
+      "learning_rate": 9.185402411208864e-06,
+      "loss": 2.6251,
       "step": 7500
     },
     {
       "epoch": 0.83,
-      "learning_rate": 9.174541110024981e-05,
-      "loss": 3.618,
+      "learning_rate": 9.174541110024981e-06,
+      "loss": 2.7169,
       "step": 7600
     },
     {
       "epoch": 0.84,
-      "learning_rate": 9.163679808841099e-05,
-      "loss": 3.7178,
+      "learning_rate": 9.1636798088411e-06,
+      "loss": 2.8072,
       "step": 7700
     },
     {
       "epoch": 0.85,
-      "learning_rate": 9.152818507657219e-05,
-      "loss": 3.6502,
+      "learning_rate": 9.152818507657219e-06,
+      "loss": 2.7628,
       "step": 7800
     },
     {
       "epoch": 0.86,
-      "learning_rate": 9.141957206473337e-05,
-      "loss": 3.6745,
+      "learning_rate": 9.141957206473336e-06,
+      "loss": 2.7874,
       "step": 7900
     },
     {
       "epoch": 0.87,
-      "learning_rate": 9.131095905289455e-05,
-      "loss": 3.6149,
+      "learning_rate": 9.131095905289454e-06,
+      "loss": 2.7073,
       "step": 8000
     },
     {
       "epoch": 0.88,
-      "learning_rate": 9.120234604105573e-05,
-      "loss": 3.5411,
+      "learning_rate": 9.120234604105573e-06,
+      "loss": 2.6771,
       "step": 8100
     },
     {
       "epoch": 0.89,
-      "learning_rate": 9.10937330292169e-05,
-      "loss": 3.4925,
+      "learning_rate": 9.109373302921692e-06,
+      "loss": 2.662,
       "step": 8200
     },
     {
       "epoch": 0.9,
-      "learning_rate": 9.098512001737809e-05,
-      "loss": 3.575,
+      "learning_rate": 9.098512001737809e-06,
+      "loss": 2.7072,
       "step": 8300
     },
     {
       "epoch": 0.91,
-      "learning_rate": 9.087650700553927e-05,
-      "loss": 3.4702,
+      "learning_rate": 9.087650700553928e-06,
+      "loss": 2.6172,
       "step": 8400
     },
     {
       "epoch": 0.92,
-      "learning_rate": 9.076789399370045e-05,
-      "loss": 3.5782,
+      "learning_rate": 9.076789399370045e-06,
+      "loss": 2.7258,
       "step": 8500
     },
     {
       "epoch": 0.93,
-      "learning_rate": 9.065928098186163e-05,
-      "loss": 3.6253,
+      "learning_rate": 9.065928098186163e-06,
+      "loss": 2.7425,
       "step": 8600
     },
     {
       "epoch": 0.94,
-      "learning_rate": 9.055066797002282e-05,
-      "loss": 3.475,
+      "learning_rate": 9.055066797002282e-06,
+      "loss": 2.628,
       "step": 8700
     },
     {
       "epoch": 0.96,
-      "learning_rate": 9.0442054958184e-05,
-      "loss": 3.5008,
+      "learning_rate": 9.0442054958184e-06,
+      "loss": 2.6762,
       "step": 8800
     },
     {
       "epoch": 0.97,
-      "learning_rate": 9.033344194634518e-05,
-      "loss": 3.5272,
+      "learning_rate": 9.033344194634518e-06,
+      "loss": 2.6539,
       "step": 8900
     },
     {
       "epoch": 0.98,
-      "learning_rate": 9.022482893450636e-05,
-      "loss": 3.5002,
+      "learning_rate": 9.022482893450635e-06,
+      "loss": 2.6896,
       "step": 9000
     },
     {
       "epoch": 0.99,
-      "learning_rate": 9.011621592266754e-05,
-      "loss": 3.611,
+      "learning_rate": 9.011621592266754e-06,
+      "loss": 2.7734,
       "step": 9100
     },
     {
       "epoch": 1.0,
-      "learning_rate": 9.000760291082873e-05,
-      "loss": 3.607,
+      "learning_rate": 9.000760291082873e-06,
+      "loss": 2.7769,
       "step": 9200
     },
     {
       "epoch": 1.01,
-      "learning_rate": 8.98989898989899e-05,
-      "loss": 3.494,
+      "learning_rate": 8.98989898989899e-06,
+      "loss": 2.7003,
       "step": 9300
     },
     {
       "epoch": 1.02,
-      "learning_rate": 8.979037688715108e-05,
-      "loss": 3.542,
+      "learning_rate": 8.97903768871511e-06,
+      "loss": 2.7512,
       "step": 9400
     },
     {
       "epoch": 1.03,
-      "learning_rate": 8.968176387531226e-05,
-      "loss": 3.468,
+      "learning_rate": 8.968176387531227e-06,
+      "loss": 2.6608,
       "step": 9500
     },
     {
       "epoch": 1.04,
-      "learning_rate": 8.957315086347344e-05,
-      "loss": 3.3716,
+      "learning_rate": 8.957315086347344e-06,
+      "loss": 2.6074,
       "step": 9600
     },
     {
       "epoch": 1.05,
-      "learning_rate": 8.946453785163463e-05,
-      "loss": 3.593,
+      "learning_rate": 8.946453785163463e-06,
+      "loss": 2.7803,
       "step": 9700
     },
     {
       "epoch": 1.06,
-      "learning_rate": 8.935592483979581e-05,
-      "loss": 3.3818,
+      "learning_rate": 8.935592483979582e-06,
+      "loss": 2.6629,
       "step": 9800
     },
     {
       "epoch": 1.08,
-      "learning_rate": 8.924731182795699e-05,
-      "loss": 3.3353,
+      "learning_rate": 8.9247311827957e-06,
+      "loss": 2.5563,
       "step": 9900
     },
     {
       "epoch": 1.09,
-      "learning_rate": 8.913869881611817e-05,
-      "loss": 3.4241,
+      "learning_rate": 8.913869881611817e-06,
+      "loss": 2.6696,
       "step": 10000
     },
     {
       "epoch": 1.1,
-      "learning_rate": 8.903008580427936e-05,
-      "loss": 3.507,
+      "learning_rate": 8.903008580427936e-06,
+      "loss": 2.7113,
       "step": 10100
     },
     {
       "epoch": 1.11,
-      "learning_rate": 8.892147279244054e-05,
-      "loss": 3.4623,
+      "learning_rate": 8.892147279244055e-06,
+      "loss": 2.7096,
       "step": 10200
     },
     {
       "epoch": 1.12,
-      "learning_rate": 8.881285978060172e-05,
-      "loss": 3.443,
+      "learning_rate": 8.881285978060172e-06,
+      "loss": 2.6906,
       "step": 10300
     },
     {
       "epoch": 1.13,
-      "learning_rate": 8.87042467687629e-05,
-      "loss": 3.4692,
+      "learning_rate": 8.870424676876291e-06,
+      "loss": 2.7112,
       "step": 10400
     },
     {
       "epoch": 1.14,
-      "learning_rate": 8.859563375692408e-05,
-      "loss": 3.424,
+      "learning_rate": 8.859563375692408e-06,
+      "loss": 2.6885,
       "step": 10500
     },
     {
       "epoch": 1.15,
-      "learning_rate": 8.848702074508526e-05,
-      "loss": 3.3139,
+      "learning_rate": 8.848702074508527e-06,
+      "loss": 2.6091,
       "step": 10600
     },
     {
       "epoch": 1.16,
-      "learning_rate": 8.837840773324644e-05,
-      "loss": 3.3319,
+      "learning_rate": 8.837840773324645e-06,
+      "loss": 2.6031,
       "step": 10700
     },
     {
       "epoch": 1.17,
-      "learning_rate": 8.826979472140762e-05,
-      "loss": 3.5363,
+      "learning_rate": 8.826979472140764e-06,
+      "loss": 2.7839,
       "step": 10800
     },
     {
       "epoch": 1.18,
-      "learning_rate": 8.81611817095688e-05,
-      "loss": 3.4598,
+      "learning_rate": 8.816118170956881e-06,
+      "loss": 2.7148,
       "step": 10900
     },
     {
       "epoch": 1.19,
-      "learning_rate": 8.805256869772998e-05,
-      "loss": 3.4883,
+      "learning_rate": 8.805256869772998e-06,
+      "loss": 2.7578,
       "step": 11000
     },
     {
       "epoch": 1.21,
-      "learning_rate": 8.794395568589118e-05,
-      "loss": 3.2751,
+      "learning_rate": 8.794395568589117e-06,
+      "loss": 2.5531,
       "step": 11100
     },
     {
       "epoch": 1.22,
-      "learning_rate": 8.783534267405236e-05,
-      "loss": 3.3685,
+      "learning_rate": 8.783534267405236e-06,
+      "loss": 2.6286,
       "step": 11200
     },
     {
       "epoch": 1.23,
-      "learning_rate": 8.772672966221354e-05,
-      "loss": 3.4111,
+      "learning_rate": 8.772672966221354e-06,
+      "loss": 2.7078,
       "step": 11300
     },
     {
       "epoch": 1.24,
-      "learning_rate": 8.761811665037472e-05,
-      "loss": 3.3984,
+      "learning_rate": 8.761811665037471e-06,
+      "loss": 2.6995,
       "step": 11400
     },
     {
       "epoch": 1.25,
-      "learning_rate": 8.750950363853591e-05,
-      "loss": 3.4246,
+      "learning_rate": 8.75095036385359e-06,
+      "loss": 2.7273,
       "step": 11500
     },
     {
       "epoch": 1.26,
-      "learning_rate": 8.740089062669709e-05,
-      "loss": 3.2751,
+      "learning_rate": 8.740089062669709e-06,
+      "loss": 2.5979,
       "step": 11600
     },
     {
       "epoch": 1.27,
-      "learning_rate": 8.729227761485827e-05,
-      "loss": 3.4855,
+      "learning_rate": 8.729227761485826e-06,
+      "loss": 2.7908,
       "step": 11700
     },
     {
       "epoch": 1.28,
-      "learning_rate": 8.718366460301945e-05,
-      "loss": 3.4712,
+      "learning_rate": 8.718366460301945e-06,
+      "loss": 2.7407,
       "step": 11800
     },
     {
       "epoch": 1.29,
-      "learning_rate": 8.707505159118062e-05,
-      "loss": 3.3221,
+      "learning_rate": 8.707505159118063e-06,
+      "loss": 2.6398,
       "step": 11900
     },
     {
       "epoch": 1.3,
-      "learning_rate": 8.696643857934181e-05,
-      "loss": 3.35,
+      "learning_rate": 8.696643857934182e-06,
+      "loss": 2.6602,
       "step": 12000
     },
     {
       "epoch": 1.31,
-      "learning_rate": 8.685782556750299e-05,
-      "loss": 3.3502,
+      "learning_rate": 8.685782556750299e-06,
+      "loss": 2.6576,
       "step": 12100
     },
     {
       "epoch": 1.33,
-      "learning_rate": 8.674921255566417e-05,
-      "loss": 3.2973,
+      "learning_rate": 8.674921255566418e-06,
+      "loss": 2.6217,
       "step": 12200
     },
     {
       "epoch": 1.34,
-      "learning_rate": 8.664059954382535e-05,
-      "loss": 3.3982,
+      "learning_rate": 8.664059954382535e-06,
+      "loss": 2.7101,
       "step": 12300
     },
     {
       "epoch": 1.35,
-      "learning_rate": 8.653198653198653e-05,
-      "loss": 3.3566,
+      "learning_rate": 8.653198653198653e-06,
+      "loss": 2.6784,
       "step": 12400
     },
     {
       "epoch": 1.36,
-      "learning_rate": 8.642337352014772e-05,
-      "loss": 3.3709,
+      "learning_rate": 8.642337352014772e-06,
+      "loss": 2.7017,
       "step": 12500
     },
     {
       "epoch": 1.37,
-      "learning_rate": 8.63147605083089e-05,
-      "loss": 3.4209,
+      "learning_rate": 8.63147605083089e-06,
+      "loss": 2.7453,
       "step": 12600
     },
     {
       "epoch": 1.38,
-      "learning_rate": 8.620614749647008e-05,
-      "loss": 3.2738,
+      "learning_rate": 8.620614749647008e-06,
+      "loss": 2.6037,
       "step": 12700
     },
     {
       "epoch": 1.39,
-      "learning_rate": 8.609753448463126e-05,
-      "loss": 3.4105,
+      "learning_rate": 8.609753448463127e-06,
+      "loss": 2.7581,
       "step": 12800
     },
     {
       "epoch": 1.4,
-      "learning_rate": 8.598892147279246e-05,
-      "loss": 3.2289,
+      "learning_rate": 8.598892147279246e-06,
+      "loss": 2.5764,
       "step": 12900
     },
     {
       "epoch": 1.41,
-      "learning_rate": 8.588030846095362e-05,
-      "loss": 3.2912,
+      "learning_rate": 8.588030846095363e-06,
+      "loss": 2.6277,
       "step": 13000
     },
     {
       "epoch": 1.42,
-      "learning_rate": 8.57716954491148e-05,
-      "loss": 3.3281,
+      "learning_rate": 8.57716954491148e-06,
+      "loss": 2.661,
       "step": 13100
     },
     {
       "epoch": 1.43,
-      "learning_rate": 8.566308243727598e-05,
-      "loss": 3.4349,
+      "learning_rate": 8.5663082437276e-06,
+      "loss": 2.8049,
       "step": 13200
     },
     {
       "epoch": 1.44,
-      "learning_rate": 8.555446942543716e-05,
-      "loss": 3.2977,
+      "learning_rate": 8.555446942543717e-06,
+      "loss": 2.6458,
       "step": 13300
     },
     {
       "epoch": 1.46,
-      "learning_rate": 8.544585641359836e-05,
-      "loss": 3.4786,
+      "learning_rate": 8.544585641359836e-06,
+      "loss": 2.8003,
       "step": 13400
     },
     {
       "epoch": 1.47,
-      "learning_rate": 8.533724340175954e-05,
-      "loss": 3.2424,
+      "learning_rate": 8.533724340175953e-06,
+      "loss": 2.5816,
       "step": 13500
     },
     {
       "epoch": 1.48,
-      "learning_rate": 8.522863038992072e-05,
-      "loss": 3.2443,
+      "learning_rate": 8.522863038992072e-06,
+      "loss": 2.6049,
       "step": 13600
     },
     {
       "epoch": 1.49,
-      "learning_rate": 8.51200173780819e-05,
-      "loss": 3.3818,
+      "learning_rate": 8.51200173780819e-06,
+      "loss": 2.727,
       "step": 13700
     },
     {
       "epoch": 1.5,
-      "learning_rate": 8.501140436624308e-05,
-      "loss": 3.3809,
+      "learning_rate": 8.501140436624309e-06,
+      "loss": 2.72,
       "step": 13800
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.490279135440427e-05,
-      "loss": 3.2922,
+      "learning_rate": 8.490279135440428e-06,
+      "loss": 2.6806,
       "step": 13900
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.479417834256545e-05,
-      "loss": 3.2152,
+      "learning_rate": 8.479417834256545e-06,
+      "loss": 2.5681,
       "step": 14000
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.468556533072663e-05,
-      "loss": 3.3144,
+      "learning_rate": 8.468556533072662e-06,
+      "loss": 2.6587,
       "step": 14100
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.457695231888781e-05,
-      "loss": 3.266,
+      "learning_rate": 8.457695231888781e-06,
+      "loss": 2.6271,
       "step": 14200
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.446833930704899e-05,
-      "loss": 3.3113,
+      "learning_rate": 8.4468339307049e-06,
+      "loss": 2.6709,
       "step": 14300
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.435972629521017e-05,
-      "loss": 3.2186,
+      "learning_rate": 8.435972629521018e-06,
+      "loss": 2.6073,
       "step": 14400
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.425111328337135e-05,
-      "loss": 3.2034,
+      "learning_rate": 8.425111328337135e-06,
+      "loss": 2.6124,
       "step": 14500
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.414250027153253e-05,
-      "loss": 3.2229,
+      "learning_rate": 8.414250027153254e-06,
+      "loss": 2.5992,
       "step": 14600
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.403388725969371e-05,
-      "loss": 3.2563,
+      "learning_rate": 8.403388725969371e-06,
+      "loss": 2.666,
       "step": 14700
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.39252742478549e-05,
-      "loss": 3.2689,
+      "learning_rate": 8.39252742478549e-06,
+      "loss": 2.6439,
       "step": 14800
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.381666123601608e-05,
-      "loss": 3.1917,
+      "learning_rate": 8.381666123601609e-06,
+      "loss": 2.5756,
       "step": 14900
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.370804822417726e-05,
-      "loss": 3.3305,
+      "learning_rate": 8.370804822417726e-06,
+      "loss": 2.7076,
       "step": 15000
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.359943521233844e-05,
-      "loss": 3.2577,
+      "learning_rate": 8.359943521233844e-06,
+      "loss": 2.6384,
       "step": 15100
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.349082220049962e-05,
-      "loss": 3.3375,
+      "learning_rate": 8.349082220049963e-06,
+      "loss": 2.728,
       "step": 15200
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.338220918866081e-05,
-      "loss": 3.2056,
+      "learning_rate": 8.338220918866082e-06,
+      "loss": 2.6078,
       "step": 15300
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.3273596176822e-05,
-      "loss": 3.2336,
+      "learning_rate": 8.327359617682199e-06,
+      "loss": 2.6352,
       "step": 15400
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.316498316498317e-05,
-      "loss": 3.2219,
+      "learning_rate": 8.316498316498316e-06,
+      "loss": 2.6383,
       "step": 15500
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.305637015314434e-05,
-      "loss": 3.3085,
+      "learning_rate": 8.305637015314435e-06,
+      "loss": 2.7244,
       "step": 15600
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.294775714130553e-05,
-      "loss": 3.2312,
+      "learning_rate": 8.294775714130554e-06,
+      "loss": 2.6331,
       "step": 15700
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.283914412946671e-05,
-      "loss": 3.2054,
+      "learning_rate": 8.283914412946672e-06,
+      "loss": 2.6084,
       "step": 15800
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.27305311176279e-05,
-      "loss": 3.3518,
+      "learning_rate": 8.27305311176279e-06,
+      "loss": 2.7714,
       "step": 15900
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.262191810578907e-05,
-      "loss": 3.22,
+      "learning_rate": 8.262191810578908e-06,
+      "loss": 2.6303,
       "step": 16000
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.251330509395025e-05,
-      "loss": 3.2564,
+      "learning_rate": 8.251330509395025e-06,
+      "loss": 2.6769,
       "step": 16100
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.240469208211145e-05,
-      "loss": 3.181,
+      "learning_rate": 8.240469208211144e-06,
+      "loss": 2.6004,
       "step": 16200
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.229607907027263e-05,
-      "loss": 3.3038,
+      "learning_rate": 8.229607907027263e-06,
+      "loss": 2.7201,
       "step": 16300
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.218746605843381e-05,
-      "loss": 3.1689,
+      "learning_rate": 8.21874660584338e-06,
+      "loss": 2.601,
       "step": 16400
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.207885304659499e-05,
-      "loss": 3.2781,
+      "learning_rate": 8.207885304659498e-06,
+      "loss": 2.6875,
       "step": 16500
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.197024003475617e-05,
-      "loss": 3.1925,
+      "learning_rate": 8.197024003475617e-06,
+      "loss": 2.6085,
       "step": 16600
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.186162702291735e-05,
-      "loss": 3.2514,
+      "learning_rate": 8.186162702291736e-06,
+      "loss": 2.6715,
       "step": 16700
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.175301401107853e-05,
-      "loss": 3.2324,
+      "learning_rate": 8.175301401107853e-06,
+      "loss": 2.6693,
       "step": 16800
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.16444009992397e-05,
-      "loss": 3.1748,
+      "learning_rate": 8.16444009992397e-06,
+      "loss": 2.6046,
       "step": 16900
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.153578798740089e-05,
-      "loss": 3.2089,
+      "learning_rate": 8.15357879874009e-06,
+      "loss": 2.6439,
       "step": 17000
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.142717497556208e-05,
-      "loss": 3.156,
+      "learning_rate": 8.142717497556209e-06,
+      "loss": 2.602,
       "step": 17100
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.131856196372326e-05,
-      "loss": 3.2313,
+      "learning_rate": 8.131856196372326e-06,
+      "loss": 2.6635,
       "step": 17200
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.120994895188444e-05,
-      "loss": 3.2406,
+      "learning_rate": 8.120994895188445e-06,
+      "loss": 2.6789,
       "step": 17300
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.110133594004562e-05,
-      "loss": 3.2273,
+      "learning_rate": 8.110133594004562e-06,
+      "loss": 2.6613,
       "step": 17400
     },
     {
       "epoch": 1.9,
-      "learning_rate": 8.09927229282068e-05,
-      "loss": 3.1781,
+      "learning_rate": 8.09927229282068e-06,
+      "loss": 2.6208,
       "step": 17500
     },
     {
       "epoch": 1.91,
-      "learning_rate": 8.088410991636799e-05,
-      "loss": 3.1407,
+      "learning_rate": 8.088410991636799e-06,
+      "loss": 2.6038,
       "step": 17600
     },
     {
       "epoch": 1.92,
-      "learning_rate": 8.077549690452917e-05,
-      "loss": 3.1266,
+      "learning_rate": 8.077549690452918e-06,
+      "loss": 2.58,
       "step": 17700
     },
     {
       "epoch": 1.93,
-      "learning_rate": 8.066688389269035e-05,
-      "loss": 3.1861,
+      "learning_rate": 8.066688389269035e-06,
+      "loss": 2.6179,
       "step": 17800
     },
     {
       "epoch": 1.94,
-      "learning_rate": 8.055827088085153e-05,
-      "loss": 3.1604,
+      "learning_rate": 8.055827088085152e-06,
+      "loss": 2.609,
       "step": 17900
     },
     {
       "epoch": 1.96,
-      "learning_rate": 8.044965786901271e-05,
-      "loss": 3.1591,
+      "learning_rate": 8.044965786901271e-06,
+      "loss": 2.6083,
       "step": 18000
     },
     {
       "epoch": 1.97,
-      "learning_rate": 8.034104485717389e-05,
-      "loss": 3.2045,
+      "learning_rate": 8.03410448571739e-06,
+      "loss": 2.6651,
       "step": 18100
     },
     {
       "epoch": 1.98,
-      "learning_rate": 8.023243184533507e-05,
-      "loss": 3.1123,
+      "learning_rate": 8.023243184533508e-06,
+      "loss": 2.5569,
       "step": 18200
     },
     {
       "epoch": 1.99,
-      "learning_rate": 8.012381883349625e-05,
-      "loss": 3.2466,
+      "learning_rate": 8.012381883349627e-06,
+      "loss": 2.7135,
       "step": 18300
     },
     {
       "epoch": 2.0,
-      "learning_rate": 8.001520582165743e-05,
-      "loss": 3.2711,
+      "learning_rate": 8.001520582165744e-06,
+      "loss": 2.7127,
       "step": 18400
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.990659280981863e-05,
-      "loss": 3.1771,
+      "learning_rate": 7.990659280981863e-06,
+      "loss": 2.667,
       "step": 18500
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.97979797979798e-05,
-      "loss": 3.1217,
+      "learning_rate": 7.97979797979798e-06,
+      "loss": 2.6272,
       "step": 18600
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.968936678614099e-05,
-      "loss": 3.0909,
+      "learning_rate": 7.9689366786141e-06,
+      "loss": 2.5742,
       "step": 18700
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.958075377430217e-05,
-      "loss": 3.0777,
+      "learning_rate": 7.958075377430217e-06,
+      "loss": 2.568,
       "step": 18800
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.947214076246335e-05,
-      "loss": 3.1697,
+      "learning_rate": 7.947214076246334e-06,
+      "loss": 2.6487,
       "step": 18900
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.936352775062454e-05,
-      "loss": 3.1588,
+      "learning_rate": 7.936352775062453e-06,
+      "loss": 2.6347,
       "step": 19000
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.925491473878572e-05,
-      "loss": 3.1334,
+      "learning_rate": 7.925491473878572e-06,
+      "loss": 2.6273,
       "step": 19100
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.91463017269469e-05,
-      "loss": 3.1442,
+      "learning_rate": 7.91463017269469e-06,
+      "loss": 2.631,
       "step": 19200
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.903768871510806e-05,
-      "loss": 3.1633,
+      "learning_rate": 7.903768871510808e-06,
+      "loss": 2.6631,
       "step": 19300
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.892907570326924e-05,
-      "loss": 3.1913,
+      "learning_rate": 7.892907570326925e-06,
+      "loss": 2.6856,
       "step": 19400
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.882046269143044e-05,
-      "loss": 3.1455,
+      "learning_rate": 7.882046269143044e-06,
+      "loss": 2.6678,
       "step": 19500
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.871184967959162e-05,
-      "loss": 3.155,
+      "learning_rate": 7.871184967959162e-06,
+      "loss": 2.6616,
       "step": 19600
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.86032366677528e-05,
-      "loss": 3.2162,
+      "learning_rate": 7.86032366677528e-06,
+      "loss": 2.694,
       "step": 19700
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.849462365591398e-05,
-      "loss": 3.1831,
+      "learning_rate": 7.849462365591398e-06,
+      "loss": 2.697,
       "step": 19800
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.838601064407516e-05,
-      "loss": 3.1012,
+      "learning_rate": 7.838601064407515e-06,
+      "loss": 2.6062,
       "step": 19900
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.827739763223635e-05,
-      "loss": 3.1624,
+      "learning_rate": 7.827739763223634e-06,
+      "loss": 2.6586,
       "step": 20000
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.816878462039753e-05,
-      "loss": 3.1208,
+      "learning_rate": 7.816878462039753e-06,
+      "loss": 2.6414,
       "step": 20100
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.806017160855871e-05,
-      "loss": 3.1124,
+      "learning_rate": 7.80601716085587e-06,
+      "loss": 2.6167,
       "step": 20200
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.795155859671989e-05,
-      "loss": 3.125,
+      "learning_rate": 7.79515585967199e-06,
+      "loss": 2.658,
       "step": 20300
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.784294558488107e-05,
-      "loss": 3.1171,
+      "learning_rate": 7.784294558488109e-06,
+      "loss": 2.6489,
       "step": 20400
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.773433257304225e-05,
-      "loss": 3.1837,
+      "learning_rate": 7.773433257304226e-06,
+      "loss": 2.7028,
       "step": 20500
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.762571956120343e-05,
-      "loss": 3.1093,
+      "learning_rate": 7.762571956120343e-06,
+      "loss": 2.6466,
       "step": 20600
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.751710654936461e-05,
-      "loss": 3.0628,
+      "learning_rate": 7.751710654936462e-06,
+      "loss": 2.5758,
       "step": 20700
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.740849353752579e-05,
-      "loss": 3.0512,
+      "learning_rate": 7.74084935375258e-06,
+      "loss": 2.5877,
       "step": 20800
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.729988052568698e-05,
-      "loss": 3.127,
+      "learning_rate": 7.729988052568699e-06,
+      "loss": 2.6322,
       "step": 20900
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.719126751384816e-05,
-      "loss": 3.1499,
+      "learning_rate": 7.719126751384816e-06,
+      "loss": 2.6729,
       "step": 21000
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.708265450200934e-05,
-      "loss": 3.069,
+      "learning_rate": 7.708265450200935e-06,
+      "loss": 2.5857,
       "step": 21100
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.697404149017052e-05,
-      "loss": 3.206,
+      "learning_rate": 7.697404149017052e-06,
+      "loss": 2.7099,
       "step": 21200
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.68654284783317e-05,
-      "loss": 3.0981,
+      "learning_rate": 7.68654284783317e-06,
+      "loss": 2.6293,
       "step": 21300
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.67568154664929e-05,
-      "loss": 3.1826,
+      "learning_rate": 7.67568154664929e-06,
+      "loss": 2.6941,
       "step": 21400
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.664820245465408e-05,
-      "loss": 3.1654,
+      "learning_rate": 7.664820245465408e-06,
+      "loss": 2.6857,
       "step": 21500
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.653958944281526e-05,
-      "loss": 3.0237,
+      "learning_rate": 7.653958944281525e-06,
+      "loss": 2.5759,
       "step": 21600
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.643097643097644e-05,
-      "loss": 3.1942,
+      "learning_rate": 7.643097643097644e-06,
+      "loss": 2.7174,
       "step": 21700
     },
     {
       "epoch": 2.37,
-      "learning_rate": 7.632236341913762e-05,
-      "loss": 3.1073,
+      "learning_rate": 7.632236341913763e-06,
+      "loss": 2.6325,
       "step": 21800
     },
     {
       "epoch": 2.38,
-      "learning_rate": 7.62137504072988e-05,
-      "loss": 3.1343,
+      "learning_rate": 7.62137504072988e-06,
+      "loss": 2.6663,
       "step": 21900
     },
     {
       "epoch": 2.39,
-      "learning_rate": 7.610513739545998e-05,
-      "loss": 3.1203,
+      "learning_rate": 7.6105137395459985e-06,
+      "loss": 2.6484,
       "step": 22000
     },
     {
       "epoch": 2.4,
-      "learning_rate": 7.599652438362116e-05,
-      "loss": 3.1676,
+      "learning_rate": 7.599652438362116e-06,
+      "loss": 2.6984,
       "step": 22100
     },
     {
       "epoch": 2.41,
-      "learning_rate": 7.588791137178234e-05,
-      "loss": 3.1682,
+      "learning_rate": 7.588791137178234e-06,
+      "loss": 2.699,
       "step": 22200
     },
     {
       "epoch": 2.42,
-      "learning_rate": 7.577929835994353e-05,
-      "loss": 3.0663,
+      "learning_rate": 7.577929835994353e-06,
+      "loss": 2.6043,
       "step": 22300
     },
     {
       "epoch": 2.43,
-      "learning_rate": 7.567068534810471e-05,
-      "loss": 3.1309,
+      "learning_rate": 7.567068534810471e-06,
+      "loss": 2.6716,
       "step": 22400
     },
     {
       "epoch": 2.44,
-      "learning_rate": 7.556207233626589e-05,
-      "loss": 3.0167,
+      "learning_rate": 7.556207233626589e-06,
+      "loss": 2.5735,
       "step": 22500
     },
     {
       "epoch": 2.45,
-      "learning_rate": 7.545345932442707e-05,
-      "loss": 3.1367,
+      "learning_rate": 7.545345932442707e-06,
+      "loss": 2.6837,
       "step": 22600
     },
     {
       "epoch": 2.47,
-      "learning_rate": 7.534484631258825e-05,
-      "loss": 3.1597,
+      "learning_rate": 7.534484631258825e-06,
+      "loss": 2.6961,
       "step": 22700
     },
     {
       "epoch": 2.48,
-      "learning_rate": 7.523623330074944e-05,
-      "loss": 3.1616,
+      "learning_rate": 7.523623330074944e-06,
+      "loss": 2.7128,
       "step": 22800
     },
     {
       "epoch": 2.49,
-      "learning_rate": 7.512762028891062e-05,
-      "loss": 3.0356,
+      "learning_rate": 7.512762028891062e-06,
+      "loss": 2.5865,
       "step": 22900
     },
     {
       "epoch": 2.5,
-      "learning_rate": 7.501900727707179e-05,
-      "loss": 3.1821,
+      "learning_rate": 7.50190072770718e-06,
+      "loss": 2.7313,
       "step": 23000
     },
     {
       "epoch": 2.51,
-      "learning_rate": 7.491039426523297e-05,
-      "loss": 3.0478,
+      "learning_rate": 7.491039426523297e-06,
+      "loss": 2.5998,
       "step": 23100
     },
     {
       "epoch": 2.52,
-      "learning_rate": 7.480178125339416e-05,
-      "loss": 3.116,
+      "learning_rate": 7.480178125339416e-06,
+      "loss": 2.6731,
       "step": 23200
     },
     {
       "epoch": 2.53,
-      "learning_rate": 7.469316824155534e-05,
-      "loss": 3.0147,
+      "learning_rate": 7.4693168241555346e-06,
+      "loss": 2.5913,
       "step": 23300
     },
     {
       "epoch": 2.54,
-      "learning_rate": 7.458455522971652e-05,
-      "loss": 3.1325,
+      "learning_rate": 7.458455522971653e-06,
+      "loss": 2.6722,
       "step": 23400
     },
     {
       "epoch": 2.55,
-      "learning_rate": 7.44759422178777e-05,
-      "loss": 2.9355,
+      "learning_rate": 7.447594221787771e-06,
+      "loss": 2.5033,
       "step": 23500
     },
     {
       "epoch": 2.56,
-      "learning_rate": 7.436732920603888e-05,
-      "loss": 3.1018,
+      "learning_rate": 7.436732920603888e-06,
+      "loss": 2.6539,
       "step": 23600
     },
     {
       "epoch": 2.57,
-      "learning_rate": 7.425871619420008e-05,
-      "loss": 3.1251,
+      "learning_rate": 7.425871619420007e-06,
+      "loss": 2.6771,
       "step": 23700
     },
     {
       "epoch": 2.58,
-      "learning_rate": 7.415010318236126e-05,
-      "loss": 3.0988,
+      "learning_rate": 7.415010318236125e-06,
+      "loss": 2.6495,
       "step": 23800
     },
     {
       "epoch": 2.6,
-      "learning_rate": 7.404149017052244e-05,
-      "loss": 3.0469,
+      "learning_rate": 7.4041490170522435e-06,
+      "loss": 2.6169,
       "step": 23900
     },
     {
       "epoch": 2.61,
-      "learning_rate": 7.393287715868362e-05,
-      "loss": 3.0617,
+      "learning_rate": 7.393287715868362e-06,
+      "loss": 2.6196,
       "step": 24000
     },
     {
       "epoch": 2.62,
-      "learning_rate": 7.38242641468448e-05,
-      "loss": 3.1484,
+      "learning_rate": 7.382426414684479e-06,
+      "loss": 2.7001,
       "step": 24100
     },
     {
       "epoch": 2.63,
-      "learning_rate": 7.371565113500597e-05,
-      "loss": 3.015,
+      "learning_rate": 7.371565113500598e-06,
+      "loss": 2.5859,
       "step": 24200
     },
     {
       "epoch": 2.64,
-      "learning_rate": 7.360703812316715e-05,
-      "loss": 3.1561,
+      "learning_rate": 7.360703812316716e-06,
+      "loss": 2.7144,
       "step": 24300
     },
     {
       "epoch": 2.65,
-      "learning_rate": 7.349842511132833e-05,
-      "loss": 3.0644,
+      "learning_rate": 7.349842511132834e-06,
+      "loss": 2.6162,
       "step": 24400
     },
     {
       "epoch": 2.66,
-      "learning_rate": 7.338981209948951e-05,
-      "loss": 3.017,
+      "learning_rate": 7.3389812099489525e-06,
+      "loss": 2.5845,
       "step": 24500
     },
     {
       "epoch": 2.67,
-      "learning_rate": 7.328119908765071e-05,
-      "loss": 3.0305,
+      "learning_rate": 7.3281199087650715e-06,
+      "loss": 2.5935,
       "step": 24600
     },
     {
       "epoch": 2.68,
-      "learning_rate": 7.317258607581189e-05,
-      "loss": 3.1057,
+      "learning_rate": 7.317258607581189e-06,
+      "loss": 2.6644,
       "step": 24700
     },
     {
       "epoch": 2.69,
-      "learning_rate": 7.306397306397307e-05,
-      "loss": 3.0648,
+      "learning_rate": 7.306397306397307e-06,
+      "loss": 2.6295,
       "step": 24800
     },
     {
       "epoch": 2.7,
-      "learning_rate": 7.295536005213425e-05,
-      "loss": 2.9583,
+      "learning_rate": 7.295536005213425e-06,
+      "loss": 2.5336,
       "step": 24900
     },
     {
       "epoch": 2.72,
-      "learning_rate": 7.284674704029543e-05,
-      "loss": 3.0673,
+      "learning_rate": 7.284674704029543e-06,
+      "loss": 2.6368,
       "step": 25000
     },
     {
       "epoch": 2.73,
-      "learning_rate": 7.273813402845662e-05,
-      "loss": 3.0524,
+      "learning_rate": 7.273813402845662e-06,
+      "loss": 2.6264,
       "step": 25100
     },
     {
       "epoch": 2.74,
-      "learning_rate": 7.26295210166178e-05,
-      "loss": 3.0098,
+      "learning_rate": 7.26295210166178e-06,
+      "loss": 2.5879,
       "step": 25200
     },
     {
       "epoch": 2.75,
-      "learning_rate": 7.252090800477898e-05,
-      "loss": 3.0,
+      "learning_rate": 7.252090800477898e-06,
+      "loss": 2.5668,
       "step": 25300
     },
     {
       "epoch": 2.76,
-      "learning_rate": 7.241229499294016e-05,
-      "loss": 3.0914,
+      "learning_rate": 7.241229499294016e-06,
+      "loss": 2.6672,
       "step": 25400
     },
     {
       "epoch": 2.77,
-      "learning_rate": 7.230368198110134e-05,
-      "loss": 3.0455,
+      "learning_rate": 7.230368198110134e-06,
+      "loss": 2.6312,
       "step": 25500
     },
     {
       "epoch": 2.78,
-      "learning_rate": 7.219506896926252e-05,
-      "loss": 3.06,
+      "learning_rate": 7.219506896926253e-06,
+      "loss": 2.6287,
       "step": 25600
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.20864559574237e-05,
-      "loss": 3.0307,
+      "learning_rate": 7.20864559574237e-06,
+      "loss": 2.6125,
       "step": 25700
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.197784294558488e-05,
-      "loss": 3.0215,
+      "learning_rate": 7.1977842945584886e-06,
+      "loss": 2.6194,
       "step": 25800
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.186922993374606e-05,
-      "loss": 3.0389,
+      "learning_rate": 7.186922993374607e-06,
+      "loss": 2.6228,
       "step": 25900
     },
     {
       "epoch": 2.82,
-      "learning_rate": 7.176061692190725e-05,
-      "loss": 3.0078,
+      "learning_rate": 7.176061692190726e-06,
+      "loss": 2.5814,
       "step": 26000
     },
     {
       "epoch": 2.83,
-      "learning_rate": 7.165200391006843e-05,
-      "loss": 2.9979,
+      "learning_rate": 7.165200391006844e-06,
+      "loss": 2.5845,
       "step": 26100
     },
     {
       "epoch": 2.85,
-      "learning_rate": 7.154339089822961e-05,
-      "loss": 3.0612,
+      "learning_rate": 7.154339089822961e-06,
+      "loss": 2.6582,
       "step": 26200
     },
     {
       "epoch": 2.86,
-      "learning_rate": 7.14347778863908e-05,
-      "loss": 3.0747,
+      "learning_rate": 7.143477788639079e-06,
+      "loss": 2.6652,
       "step": 26300
     },
     {
       "epoch": 2.87,
-      "learning_rate": 7.132616487455197e-05,
-      "loss": 3.0439,
+      "learning_rate": 7.1326164874551975e-06,
+      "loss": 2.6307,
       "step": 26400
     },
     {
       "epoch": 2.88,
-      "learning_rate": 7.121755186271317e-05,
-      "loss": 3.0511,
+      "learning_rate": 7.1217551862713165e-06,
+      "loss": 2.6338,
       "step": 26500
     },
     {
       "epoch": 2.89,
-      "learning_rate": 7.110893885087435e-05,
-      "loss": 3.0268,
+      "learning_rate": 7.110893885087435e-06,
+      "loss": 2.6084,
       "step": 26600
     },
     {
       "epoch": 2.9,
-      "learning_rate": 7.100032583903551e-05,
-      "loss": 3.0712,
+      "learning_rate": 7.100032583903552e-06,
+      "loss": 2.6638,
       "step": 26700
     },
     {
       "epoch": 2.91,
-      "learning_rate": 7.089171282719669e-05,
-      "loss": 3.0466,
+      "learning_rate": 7.08917128271967e-06,
+      "loss": 2.6421,
       "step": 26800
     },
     {
       "epoch": 2.92,
-      "learning_rate": 7.078309981535787e-05,
-      "loss": 3.0257,
+      "learning_rate": 7.078309981535788e-06,
+      "loss": 2.6346,
       "step": 26900
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.067448680351907e-05,
-      "loss": 3.0384,
+      "learning_rate": 7.067448680351907e-06,
+      "loss": 2.6266,
       "step": 27000
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.056587379168025e-05,
-      "loss": 2.9557,
+      "learning_rate": 7.056587379168025e-06,
+      "loss": 2.5764,
       "step": 27100
     },
     {
       "epoch": 2.95,
-      "learning_rate": 7.045726077984143e-05,
-      "loss": 3.0617,
+      "learning_rate": 7.045726077984143e-06,
+      "loss": 2.6617,
       "step": 27200
     },
     {
       "epoch": 2.97,
-      "learning_rate": 7.03486477680026e-05,
-      "loss": 3.0958,
+      "learning_rate": 7.034864776800261e-06,
+      "loss": 2.6819,
       "step": 27300
     },
     {
       "epoch": 2.98,
-      "learning_rate": 7.02400347561638e-05,
-      "loss": 3.0421,
+      "learning_rate": 7.02400347561638e-06,
+      "loss": 2.6382,
       "step": 27400
     },
     {
       "epoch": 2.99,
-      "learning_rate": 7.013142174432498e-05,
-      "loss": 3.0083,
+      "learning_rate": 7.013142174432498e-06,
+      "loss": 2.6016,
       "step": 27500
     },
     {
       "epoch": 3.0,
-      "learning_rate": 7.002280873248616e-05,
-      "loss": 3.0199,
+      "learning_rate": 7.0022808732486154e-06,
+      "loss": 2.62,
       "step": 27600
     },
     {
       "epoch": 3.01,
-      "learning_rate": 6.991419572064734e-05,
-      "loss": 3.0205,
+      "learning_rate": 6.991419572064734e-06,
+      "loss": 2.6396,
       "step": 27700
     },
     {
       "epoch": 3.02,
-      "learning_rate": 6.980558270880852e-05,
-      "loss": 3.0701,
+      "learning_rate": 6.980558270880852e-06,
+      "loss": 2.6956,
       "step": 27800
     },
     {
       "epoch": 3.03,
-      "learning_rate": 6.96969696969697e-05,
-      "loss": 2.8957,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 2.5377,
       "step": 27900
     },
     {
       "epoch": 3.04,
-      "learning_rate": 6.958835668513088e-05,
-      "loss": 2.9442,
+      "learning_rate": 6.958835668513089e-06,
+      "loss": 2.569,
       "step": 28000
     },
     {
       "epoch": 3.05,
-      "learning_rate": 6.947974367329206e-05,
-      "loss": 2.9768,
+      "learning_rate": 6.947974367329206e-06,
+      "loss": 2.5924,
       "step": 28100
     },
     {
       "epoch": 3.06,
-      "learning_rate": 6.937113066145324e-05,
-      "loss": 3.0057,
+      "learning_rate": 6.937113066145324e-06,
+      "loss": 2.6204,
       "step": 28200
     },
     {
       "epoch": 3.07,
-      "learning_rate": 6.926251764961442e-05,
-      "loss": 2.9816,
+      "learning_rate": 6.9262517649614425e-06,
+      "loss": 2.6171,
       "step": 28300
     },
     {
       "epoch": 3.08,
-      "learning_rate": 6.915390463777561e-05,
-      "loss": 2.9866,
+      "learning_rate": 6.9153904637775616e-06,
+      "loss": 2.6091,
       "step": 28400
     },
     {
       "epoch": 3.1,
-      "learning_rate": 6.904529162593679e-05,
-      "loss": 3.056,
+      "learning_rate": 6.90452916259368e-06,
+      "loss": 2.6877,
       "step": 28500
     },
     {
       "epoch": 3.11,
-      "learning_rate": 6.893667861409797e-05,
-      "loss": 2.9423,
+      "learning_rate": 6.893667861409797e-06,
+      "loss": 2.5785,
       "step": 28600
     },
     {
       "epoch": 3.12,
-      "learning_rate": 6.882806560225915e-05,
-      "loss": 2.9433,
+      "learning_rate": 6.882806560225915e-06,
+      "loss": 2.587,
       "step": 28700
     },
     {
       "epoch": 3.13,
-      "learning_rate": 6.871945259042033e-05,
-      "loss": 2.9175,
+      "learning_rate": 6.871945259042033e-06,
+      "loss": 2.5562,
       "step": 28800
     },
     {
       "epoch": 3.14,
-      "learning_rate": 6.861083957858153e-05,
-      "loss": 3.0126,
+      "learning_rate": 6.861083957858152e-06,
+      "loss": 2.6319,
       "step": 28900
     },
     {
       "epoch": 3.15,
-      "learning_rate": 6.85022265667427e-05,
-      "loss": 2.9797,
+      "learning_rate": 6.8502226566742705e-06,
+      "loss": 2.607,
       "step": 29000
     },
     {
       "epoch": 3.16,
-      "learning_rate": 6.839361355490389e-05,
-      "loss": 2.9233,
+      "learning_rate": 6.839361355490388e-06,
+      "loss": 2.5583,
       "step": 29100
     },
     {
       "epoch": 3.17,
-      "learning_rate": 6.828500054306506e-05,
-      "loss": 2.953,
+      "learning_rate": 6.828500054306506e-06,
+      "loss": 2.6092,
       "step": 29200
     },
     {
       "epoch": 3.18,
-      "learning_rate": 6.817638753122624e-05,
-      "loss": 2.9087,
+      "learning_rate": 6.817638753122625e-06,
+      "loss": 2.5571,
       "step": 29300
     },
     {
       "epoch": 3.19,
-      "learning_rate": 6.806777451938742e-05,
-      "loss": 2.951,
+      "learning_rate": 6.806777451938743e-06,
+      "loss": 2.5965,
       "step": 29400
     },
     {
       "epoch": 3.2,
-      "learning_rate": 6.79591615075486e-05,
-      "loss": 3.0381,
+      "learning_rate": 6.795916150754861e-06,
+      "loss": 2.6739,
       "step": 29500
     },
     {
       "epoch": 3.21,
-      "learning_rate": 6.785054849570978e-05,
-      "loss": 2.9996,
+      "learning_rate": 6.785054849570979e-06,
+      "loss": 2.6389,
       "step": 29600
     },
     {
       "epoch": 3.23,
-      "learning_rate": 6.774193548387096e-05,
-      "loss": 2.9855,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 2.6198,
       "step": 29700
     },
     {
       "epoch": 3.24,
-      "learning_rate": 6.763332247203216e-05,
-      "loss": 2.9895,
+      "learning_rate": 6.763332247203216e-06,
+      "loss": 2.6363,
       "step": 29800
     },
     {
       "epoch": 3.25,
-      "learning_rate": 6.752470946019334e-05,
-      "loss": 2.9553,
+      "learning_rate": 6.752470946019334e-06,
+      "loss": 2.5963,
       "step": 29900
     },
     {
       "epoch": 3.26,
-      "learning_rate": 6.741609644835452e-05,
-      "loss": 3.0319,
+      "learning_rate": 6.741609644835452e-06,
+      "loss": 2.6672,
       "step": 30000
     },
     {
       "epoch": 3.27,
-      "learning_rate": 6.73074834365157e-05,
-      "loss": 3.0371,
+      "learning_rate": 6.730748343651569e-06,
+      "loss": 2.6811,
       "step": 30100
     },
     {
       "epoch": 3.28,
-      "learning_rate": 6.719887042467688e-05,
-      "loss": 2.9884,
+      "learning_rate": 6.719887042467688e-06,
+      "loss": 2.6339,
       "step": 30200
     },
     {
       "epoch": 3.29,
-      "learning_rate": 6.709025741283807e-05,
-      "loss": 2.9377,
+      "learning_rate": 6.709025741283807e-06,
+      "loss": 2.5835,
       "step": 30300
     },
     {
       "epoch": 3.3,
-      "learning_rate": 6.698164440099924e-05,
-      "loss": 2.9495,
+      "learning_rate": 6.698164440099925e-06,
+      "loss": 2.5971,
       "step": 30400
     },
     {
       "epoch": 3.31,
-      "learning_rate": 6.687303138916042e-05,
-      "loss": 2.9938,
+      "learning_rate": 6.687303138916043e-06,
+      "loss": 2.6315,
       "step": 30500
     },
     {
       "epoch": 3.32,
-      "learning_rate": 6.67644183773216e-05,
-      "loss": 2.9422,
+      "learning_rate": 6.67644183773216e-06,
+      "loss": 2.5878,
       "step": 30600
     },
     {
       "epoch": 3.33,
-      "learning_rate": 6.665580536548279e-05,
-      "loss": 3.0207,
+      "learning_rate": 6.665580536548279e-06,
+      "loss": 2.6581,
       "step": 30700
     },
     {
       "epoch": 3.35,
-      "learning_rate": 6.654719235364397e-05,
-      "loss": 2.8544,
+      "learning_rate": 6.654719235364397e-06,
+      "loss": 2.5079,
       "step": 30800
     },
     {
       "epoch": 3.36,
-      "learning_rate": 6.643857934180515e-05,
-      "loss": 2.9468,
+      "learning_rate": 6.6438579341805155e-06,
+      "loss": 2.5946,
       "step": 30900
     },
     {
       "epoch": 3.37,
-      "learning_rate": 6.632996632996633e-05,
-      "loss": 2.9787,
+      "learning_rate": 6.632996632996634e-06,
+      "loss": 2.6383,
       "step": 31000
     },
     {
       "epoch": 3.38,
-      "learning_rate": 6.622135331812751e-05,
-      "loss": 2.9319,
+      "learning_rate": 6.622135331812751e-06,
+      "loss": 2.5752,
       "step": 31100
     },
     {
       "epoch": 3.39,
-      "learning_rate": 6.61127403062887e-05,
-      "loss": 2.9638,
+      "learning_rate": 6.61127403062887e-06,
+      "loss": 2.61,
       "step": 31200
     },
     {
       "epoch": 3.4,
-      "learning_rate": 6.600412729444988e-05,
-      "loss": 2.9358,
+      "learning_rate": 6.600412729444988e-06,
+      "loss": 2.575,
       "step": 31300
     },
     {
       "epoch": 3.41,
-      "learning_rate": 6.589551428261106e-05,
-      "loss": 2.979,
+      "learning_rate": 6.589551428261106e-06,
+      "loss": 2.6239,
       "step": 31400
     },
     {
       "epoch": 3.42,
-      "learning_rate": 6.578690127077224e-05,
-      "loss": 2.9887,
+      "learning_rate": 6.5786901270772245e-06,
+      "loss": 2.6439,
       "step": 31500
     },
     {
       "epoch": 3.43,
-      "learning_rate": 6.567828825893342e-05,
-      "loss": 2.8997,
+      "learning_rate": 6.567828825893342e-06,
+      "loss": 2.5511,
       "step": 31600
     },
     {
       "epoch": 3.44,
-      "learning_rate": 6.55696752470946e-05,
-      "loss": 2.9228,
+      "learning_rate": 6.556967524709461e-06,
+      "loss": 2.5801,
       "step": 31700
     },
     {
       "epoch": 3.45,
-      "learning_rate": 6.546106223525578e-05,
-      "loss": 2.9675,
+      "learning_rate": 6.546106223525579e-06,
+      "loss": 2.6397,
       "step": 31800
     },
     {
       "epoch": 3.46,
-      "learning_rate": 6.535244922341696e-05,
-      "loss": 2.8826,
+      "learning_rate": 6.535244922341697e-06,
+      "loss": 2.5415,
       "step": 31900
     },
     {
       "epoch": 3.48,
-      "learning_rate": 6.524383621157814e-05,
-      "loss": 3.0264,
+      "learning_rate": 6.5243836211578145e-06,
+      "loss": 2.6778,
       "step": 32000
     },
     {
       "epoch": 3.49,
-      "learning_rate": 6.513522319973934e-05,
-      "loss": 2.9325,
+      "learning_rate": 6.513522319973934e-06,
+      "loss": 2.6048,
       "step": 32100
     },
     {
       "epoch": 3.5,
-      "learning_rate": 6.502661018790052e-05,
-      "loss": 2.9294,
+      "learning_rate": 6.502661018790052e-06,
+      "loss": 2.5939,
       "step": 32200
     },
     {
       "epoch": 3.51,
-      "learning_rate": 6.49179971760617e-05,
-      "loss": 2.9156,
+      "learning_rate": 6.49179971760617e-06,
+      "loss": 2.5753,
       "step": 32300
     },
     {
       "epoch": 3.52,
-      "learning_rate": 6.480938416422288e-05,
-      "loss": 2.9291,
+      "learning_rate": 6.480938416422288e-06,
+      "loss": 2.5881,
       "step": 32400
     },
     {
       "epoch": 3.53,
-      "learning_rate": 6.470077115238406e-05,
-      "loss": 2.9219,
+      "learning_rate": 6.470077115238405e-06,
+      "loss": 2.5772,
       "step": 32500
     },
     {
       "epoch": 3.54,
-      "learning_rate": 6.459215814054525e-05,
-      "loss": 3.0269,
+      "learning_rate": 6.459215814054525e-06,
+      "loss": 2.6937,
       "step": 32600
     },
     {
       "epoch": 3.55,
-      "learning_rate": 6.448354512870643e-05,
-      "loss": 3.0573,
+      "learning_rate": 6.448354512870642e-06,
+      "loss": 2.7197,
       "step": 32700
     },
     {
       "epoch": 3.56,
-      "learning_rate": 6.437493211686761e-05,
-      "loss": 2.979,
+      "learning_rate": 6.437493211686761e-06,
+      "loss": 2.6423,
       "step": 32800
     },
     {
       "epoch": 3.57,
-      "learning_rate": 6.426631910502879e-05,
-      "loss": 2.9678,
+      "learning_rate": 6.426631910502879e-06,
+      "loss": 2.6384,
       "step": 32900
     },
     {
       "epoch": 3.58,
-      "learning_rate": 6.415770609318996e-05,
-      "loss": 2.8691,
+      "learning_rate": 6.415770609318996e-06,
+      "loss": 2.5382,
       "step": 33000
     },
     {
       "epoch": 3.6,
-      "learning_rate": 6.404909308135115e-05,
-      "loss": 3.0819,
+      "learning_rate": 6.404909308135115e-06,
+      "loss": 2.7512,
       "step": 33100
     },
     {
       "epoch": 3.61,
-      "learning_rate": 6.394048006951233e-05,
-      "loss": 2.9761,
+      "learning_rate": 6.394048006951233e-06,
+      "loss": 2.6328,
       "step": 33200
     },
     {
       "epoch": 3.62,
-      "learning_rate": 6.383186705767351e-05,
-      "loss": 3.0444,
+      "learning_rate": 6.383186705767351e-06,
+      "loss": 2.7133,
       "step": 33300
     },
     {
       "epoch": 3.63,
-      "learning_rate": 6.372325404583469e-05,
-      "loss": 2.91,
+      "learning_rate": 6.3723254045834695e-06,
+      "loss": 2.5745,
       "step": 33400
     },
     {
       "epoch": 3.64,
-      "learning_rate": 6.361464103399588e-05,
-      "loss": 2.9845,
+      "learning_rate": 6.3614641033995885e-06,
+      "loss": 2.648,
       "step": 33500
     },
     {
       "epoch": 3.65,
-      "learning_rate": 6.350602802215706e-05,
-      "loss": 2.9298,
+      "learning_rate": 6.350602802215706e-06,
+      "loss": 2.5938,
       "step": 33600
     },
     {
       "epoch": 3.66,
-      "learning_rate": 6.339741501031824e-05,
-      "loss": 2.9518,
+      "learning_rate": 6.339741501031824e-06,
+      "loss": 2.6273,
       "step": 33700
     },
     {
       "epoch": 3.67,
-      "learning_rate": 6.328880199847942e-05,
-      "loss": 2.9359,
+      "learning_rate": 6.328880199847942e-06,
+      "loss": 2.6032,
       "step": 33800
     },
     {
       "epoch": 3.68,
-      "learning_rate": 6.31801889866406e-05,
-      "loss": 2.9675,
+      "learning_rate": 6.31801889866406e-06,
+      "loss": 2.6407,
       "step": 33900
     },
     {
       "epoch": 3.69,
-      "learning_rate": 6.30715759748018e-05,
-      "loss": 2.8137,
+      "learning_rate": 6.307157597480179e-06,
+      "loss": 2.4937,
       "step": 34000
     },
     {
       "epoch": 3.7,
-      "learning_rate": 6.296296296296296e-05,
-      "loss": 3.0078,
+      "learning_rate": 6.296296296296297e-06,
+      "loss": 2.6804,
       "step": 34100
     },
     {
       "epoch": 3.71,
-      "learning_rate": 6.285434995112414e-05,
-      "loss": 3.0557,
+      "learning_rate": 6.285434995112415e-06,
+      "loss": 2.7139,
       "step": 34200
     },
     {
       "epoch": 3.73,
-      "learning_rate": 6.274573693928532e-05,
-      "loss": 2.9161,
+      "learning_rate": 6.274573693928533e-06,
+      "loss": 2.5752,
       "step": 34300
     },
     {
       "epoch": 3.74,
-      "learning_rate": 6.26371239274465e-05,
-      "loss": 2.9669,
+      "learning_rate": 6.263712392744651e-06,
+      "loss": 2.6275,
       "step": 34400
     },
     {
       "epoch": 3.75,
-      "learning_rate": 6.25285109156077e-05,
-      "loss": 2.9938,
+      "learning_rate": 6.25285109156077e-06,
+      "loss": 2.6574,
       "step": 34500
     },
     {
       "epoch": 3.76,
-      "learning_rate": 6.241989790376887e-05,
-      "loss": 2.9183,
+      "learning_rate": 6.2419897903768875e-06,
+      "loss": 2.5934,
       "step": 34600
     },
     {
       "epoch": 3.77,
-      "learning_rate": 6.231128489193005e-05,
-      "loss": 2.9855,
+      "learning_rate": 6.231128489193006e-06,
+      "loss": 2.6603,
       "step": 34700
     },
     {
       "epoch": 3.78,
-      "learning_rate": 6.220267188009123e-05,
-      "loss": 2.8912,
+      "learning_rate": 6.220267188009124e-06,
+      "loss": 2.57,
       "step": 34800
     },
     {
       "epoch": 3.79,
-      "learning_rate": 6.209405886825243e-05,
-      "loss": 2.8896,
+      "learning_rate": 6.209405886825243e-06,
+      "loss": 2.576,
       "step": 34900
     },
     {
       "epoch": 3.8,
-      "learning_rate": 6.198544585641361e-05,
-      "loss": 2.953,
+      "learning_rate": 6.198544585641361e-06,
+      "loss": 2.623,
       "step": 35000
     },
     {
       "epoch": 3.81,
-      "learning_rate": 6.187683284457479e-05,
-      "loss": 2.8615,
+      "learning_rate": 6.187683284457478e-06,
+      "loss": 2.5484,
       "step": 35100
     },
     {
       "epoch": 3.82,
-      "learning_rate": 6.176821983273597e-05,
-      "loss": 2.9473,
+      "learning_rate": 6.176821983273596e-06,
+      "loss": 2.625,
       "step": 35200
     },
     {
       "epoch": 3.83,
-      "learning_rate": 6.165960682089715e-05,
-      "loss": 2.9928,
+      "learning_rate": 6.1659606820897146e-06,
+      "loss": 2.6665,
       "step": 35300
     },
     {
       "epoch": 3.84,
-      "learning_rate": 6.155099380905833e-05,
-      "loss": 2.8082,
+      "learning_rate": 6.155099380905834e-06,
+      "loss": 2.4996,
       "step": 35400
     },
     {
       "epoch": 3.86,
-      "learning_rate": 6.144238079721951e-05,
-      "loss": 2.9584,
+      "learning_rate": 6.144238079721952e-06,
+      "loss": 2.6436,
       "step": 35500
     },
     {
       "epoch": 3.87,
-      "learning_rate": 6.133376778538069e-05,
-      "loss": 2.9186,
+      "learning_rate": 6.133376778538069e-06,
+      "loss": 2.6167,
       "step": 35600
     },
     {
       "epoch": 3.88,
-      "learning_rate": 6.122515477354187e-05,
-      "loss": 2.9403,
+      "learning_rate": 6.122515477354187e-06,
+      "loss": 2.622,
       "step": 35700
     },
     {
       "epoch": 3.89,
-      "learning_rate": 6.111654176170305e-05,
-      "loss": 2.8996,
+      "learning_rate": 6.111654176170305e-06,
+      "loss": 2.6065,
       "step": 35800
     },
     {
       "epoch": 3.9,
-      "learning_rate": 6.100792874986424e-05,
-      "loss": 3.0083,
+      "learning_rate": 6.100792874986424e-06,
+      "loss": 2.6941,
       "step": 35900
     },
     {
       "epoch": 3.91,
-      "learning_rate": 6.089931573802542e-05,
-      "loss": 2.9722,
+      "learning_rate": 6.0899315738025425e-06,
+      "loss": 2.6445,
       "step": 36000
     },
     {
       "epoch": 3.92,
-      "learning_rate": 6.07907027261866e-05,
-      "loss": 2.8405,
+      "learning_rate": 6.07907027261866e-06,
+      "loss": 2.5203,
       "step": 36100
     },
     {
       "epoch": 3.93,
-      "learning_rate": 6.068208971434778e-05,
-      "loss": 2.971,
+      "learning_rate": 6.068208971434778e-06,
+      "loss": 2.6714,
       "step": 36200
     },
     {
       "epoch": 3.94,
-      "learning_rate": 6.057347670250897e-05,
-      "loss": 2.9748,
+      "learning_rate": 6.057347670250897e-06,
+      "loss": 2.6593,
       "step": 36300
     },
     {
       "epoch": 3.95,
-      "learning_rate": 6.046486369067015e-05,
-      "loss": 2.8846,
+      "learning_rate": 6.046486369067015e-06,
+      "loss": 2.5758,
       "step": 36400
     },
     {
       "epoch": 3.96,
-      "learning_rate": 6.0356250678831327e-05,
-      "loss": 2.8576,
+      "learning_rate": 6.035625067883133e-06,
+      "loss": 2.5604,
       "step": 36500
     },
     {
       "epoch": 3.98,
-      "learning_rate": 6.0247637666992506e-05,
-      "loss": 2.9272,
+      "learning_rate": 6.024763766699251e-06,
+      "loss": 2.6092,
       "step": 36600
     },
     {
       "epoch": 3.99,
-      "learning_rate": 6.0139024655153686e-05,
-      "loss": 2.9479,
+      "learning_rate": 6.013902465515369e-06,
+      "loss": 2.6276,
       "step": 36700
     },
     {
       "epoch": 4.0,
-      "learning_rate": 6.003041164331488e-05,
-      "loss": 2.8707,
+      "learning_rate": 6.003041164331488e-06,
+      "loss": 2.5607,
       "step": 36800
     },
     {
       "epoch": 4.01,
-      "learning_rate": 5.992179863147605e-05,
-      "loss": 2.9569,
+      "learning_rate": 5.992179863147606e-06,
+      "loss": 2.6589,
       "step": 36900
     },
     {
       "epoch": 4.02,
-      "learning_rate": 5.981318561963723e-05,
-      "loss": 2.8839,
+      "learning_rate": 5.981318561963724e-06,
+      "loss": 2.605,
       "step": 37000
     },
     {
       "epoch": 4.03,
-      "learning_rate": 5.970457260779841e-05,
-      "loss": 2.9383,
+      "learning_rate": 5.9704572607798414e-06,
+      "loss": 2.6398,
       "step": 37100
     },
     {
       "epoch": 4.04,
-      "learning_rate": 5.959595959595959e-05,
-      "loss": 2.9082,
+      "learning_rate": 5.95959595959596e-06,
+      "loss": 2.6199,
       "step": 37200
     },
     {
       "epoch": 4.05,
-      "learning_rate": 5.9487346584120786e-05,
-      "loss": 2.8685,
+      "learning_rate": 5.948734658412079e-06,
+      "loss": 2.5676,
       "step": 37300
     },
     {
       "epoch": 4.06,
-      "learning_rate": 5.9378733572281966e-05,
-      "loss": 2.8974,
+      "learning_rate": 5.937873357228197e-06,
+      "loss": 2.6085,
       "step": 37400
     },
     {
       "epoch": 4.07,
-      "learning_rate": 5.9270120560443146e-05,
-      "loss": 2.9803,
+      "learning_rate": 5.927012056044315e-06,
+      "loss": 2.6917,
       "step": 37500
     },
     {
       "epoch": 4.08,
-      "learning_rate": 5.916150754860432e-05,
-      "loss": 2.8538,
+      "learning_rate": 5.916150754860432e-06,
+      "loss": 2.5773,
       "step": 37600
     },
     {
       "epoch": 4.09,
-      "learning_rate": 5.905289453676551e-05,
-      "loss": 2.7821,
+      "learning_rate": 5.905289453676551e-06,
+      "loss": 2.506,
       "step": 37700
     },
     {
       "epoch": 4.11,
-      "learning_rate": 5.894428152492669e-05,
-      "loss": 2.8806,
+      "learning_rate": 5.894428152492669e-06,
+      "loss": 2.6047,
       "step": 37800
     },
     {
       "epoch": 4.12,
-      "learning_rate": 5.883566851308787e-05,
-      "loss": 2.871,
+      "learning_rate": 5.8835668513087876e-06,
+      "loss": 2.5927,
       "step": 37900
     },
     {
       "epoch": 4.13,
-      "learning_rate": 5.872705550124905e-05,
-      "loss": 2.954,
+      "learning_rate": 5.872705550124905e-06,
+      "loss": 2.6702,
       "step": 38000
     },
     {
       "epoch": 4.14,
-      "learning_rate": 5.861844248941023e-05,
-      "loss": 2.8033,
+      "learning_rate": 5.861844248941023e-06,
+      "loss": 2.5303,
       "step": 38100
     },
     {
       "epoch": 4.15,
-      "learning_rate": 5.850982947757142e-05,
-      "loss": 2.8521,
+      "learning_rate": 5.850982947757142e-06,
+      "loss": 2.5736,
       "step": 38200
     },
     {
       "epoch": 4.16,
-      "learning_rate": 5.84012164657326e-05,
-      "loss": 2.7813,
+      "learning_rate": 5.84012164657326e-06,
+      "loss": 2.5033,
       "step": 38300
     },
     {
       "epoch": 4.17,
-      "learning_rate": 5.829260345389378e-05,
-      "loss": 2.823,
+      "learning_rate": 5.829260345389378e-06,
+      "loss": 2.5523,
       "step": 38400
     },
     {
       "epoch": 4.18,
-      "learning_rate": 5.818399044205496e-05,
-      "loss": 2.862,
+      "learning_rate": 5.818399044205496e-06,
+      "loss": 2.5858,
       "step": 38500
     },
     {
       "epoch": 4.19,
-      "learning_rate": 5.807537743021614e-05,
-      "loss": 2.8487,
+      "learning_rate": 5.807537743021614e-06,
+      "loss": 2.573,
       "step": 38600
     },
     {
       "epoch": 4.2,
-      "learning_rate": 5.7966764418377325e-05,
-      "loss": 2.8546,
+      "learning_rate": 5.796676441837733e-06,
+      "loss": 2.5753,
       "step": 38700
     },
     {
       "epoch": 4.21,
-      "learning_rate": 5.7858151406538505e-05,
-      "loss": 2.9429,
+      "learning_rate": 5.785815140653851e-06,
+      "loss": 2.6649,
       "step": 38800
     },
     {
       "epoch": 4.23,
-      "learning_rate": 5.7749538394699685e-05,
-      "loss": 2.9066,
+      "learning_rate": 5.774953839469969e-06,
+      "loss": 2.6291,
       "step": 38900
     },
     {
       "epoch": 4.24,
-      "learning_rate": 5.7640925382860865e-05,
-      "loss": 3.0249,
+      "learning_rate": 5.7640925382860865e-06,
+      "loss": 2.7335,
       "step": 39000
     },
     {
       "epoch": 4.25,
-      "learning_rate": 5.7532312371022045e-05,
-      "loss": 2.8215,
+      "learning_rate": 5.753231237102205e-06,
+      "loss": 2.5413,
       "step": 39100
     },
     {
       "epoch": 4.26,
-      "learning_rate": 5.742369935918324e-05,
-      "loss": 2.8526,
+      "learning_rate": 5.742369935918324e-06,
+      "loss": 2.5735,
       "step": 39200
     },
     {
       "epoch": 4.27,
-      "learning_rate": 5.731508634734442e-05,
-      "loss": 2.9065,
+      "learning_rate": 5.731508634734442e-06,
+      "loss": 2.628,
       "step": 39300
     },
     {
       "epoch": 4.28,
-      "learning_rate": 5.72064733355056e-05,
-      "loss": 2.8976,
+      "learning_rate": 5.72064733355056e-06,
+      "loss": 2.6272,
       "step": 39400
     },
     {
       "epoch": 4.29,
-      "learning_rate": 5.709786032366677e-05,
-      "loss": 2.9547,
+      "learning_rate": 5.709786032366677e-06,
+      "loss": 2.6771,
       "step": 39500
     },
     {
       "epoch": 4.3,
-      "learning_rate": 5.6989247311827965e-05,
-      "loss": 2.9062,
+      "learning_rate": 5.698924731182796e-06,
+      "loss": 2.6215,
       "step": 39600
     },
     {
       "epoch": 4.31,
-      "learning_rate": 5.6880634299989144e-05,
-      "loss": 2.8824,
+      "learning_rate": 5.6880634299989144e-06,
+      "loss": 2.6103,
       "step": 39700
     },
     {
       "epoch": 4.32,
-      "learning_rate": 5.6772021288150324e-05,
-      "loss": 2.8768,
+      "learning_rate": 5.677202128815033e-06,
+      "loss": 2.6128,
       "step": 39800
     },
     {
       "epoch": 4.33,
-      "learning_rate": 5.6663408276311504e-05,
-      "loss": 2.8265,
+      "learning_rate": 5.666340827631151e-06,
+      "loss": 2.568,
       "step": 39900
     },
     {
       "epoch": 4.34,
-      "learning_rate": 5.6554795264472684e-05,
-      "loss": 2.9223,
+      "learning_rate": 5.655479526447268e-06,
+      "loss": 2.654,
       "step": 40000
     },
     {
       "epoch": 4.36,
-      "learning_rate": 5.644618225263387e-05,
-      "loss": 2.8611,
+      "learning_rate": 5.644618225263387e-06,
+      "loss": 2.5993,
       "step": 40100
     },
     {
       "epoch": 4.37,
-      "learning_rate": 5.633756924079505e-05,
-      "loss": 2.8157,
+      "learning_rate": 5.633756924079505e-06,
+      "loss": 2.5388,
       "step": 40200
     },
     {
       "epoch": 4.38,
-      "learning_rate": 5.622895622895623e-05,
-      "loss": 2.848,
+      "learning_rate": 5.622895622895623e-06,
+      "loss": 2.5835,
       "step": 40300
     },
     {
       "epoch": 4.39,
-      "learning_rate": 5.612034321711741e-05,
-      "loss": 2.8369,
+      "learning_rate": 5.6120343217117416e-06,
+      "loss": 2.5612,
       "step": 40400
     },
     {
       "epoch": 4.4,
-      "learning_rate": 5.601173020527859e-05,
-      "loss": 2.966,
+      "learning_rate": 5.601173020527859e-06,
+      "loss": 2.6974,
       "step": 40500
     },
     {
       "epoch": 4.41,
-      "learning_rate": 5.590311719343978e-05,
-      "loss": 2.9007,
+      "learning_rate": 5.590311719343978e-06,
+      "loss": 2.6306,
       "step": 40600
     },
     {
       "epoch": 4.42,
-      "learning_rate": 5.579450418160096e-05,
-      "loss": 2.9092,
+      "learning_rate": 5.579450418160096e-06,
+      "loss": 2.6444,
       "step": 40700
     },
     {
       "epoch": 4.43,
-      "learning_rate": 5.568589116976214e-05,
-      "loss": 2.911,
+      "learning_rate": 5.568589116976214e-06,
+      "loss": 2.6471,
       "step": 40800
     },
     {
       "epoch": 4.44,
-      "learning_rate": 5.557727815792332e-05,
-      "loss": 2.8352,
+      "learning_rate": 5.557727815792332e-06,
+      "loss": 2.5671,
       "step": 40900
     },
     {
       "epoch": 4.45,
-      "learning_rate": 5.546866514608451e-05,
-      "loss": 2.814,
+      "learning_rate": 5.546866514608451e-06,
+      "loss": 2.5424,
       "step": 41000
     },
     {
       "epoch": 4.46,
-      "learning_rate": 5.536005213424569e-05,
-      "loss": 2.8483,
+      "learning_rate": 5.536005213424569e-06,
+      "loss": 2.5915,
       "step": 41100
     },
     {
       "epoch": 4.47,
-      "learning_rate": 5.525143912240687e-05,
-      "loss": 2.7428,
+      "learning_rate": 5.525143912240687e-06,
+      "loss": 2.4773,
       "step": 41200
     },
     {
       "epoch": 4.49,
-      "learning_rate": 5.514282611056805e-05,
-      "loss": 2.7516,
+      "learning_rate": 5.514282611056805e-06,
+      "loss": 2.4958,
       "step": 41300
     },
     {
       "epoch": 4.5,
-      "learning_rate": 5.503421309872922e-05,
-      "loss": 2.9075,
+      "learning_rate": 5.503421309872923e-06,
+      "loss": 2.6378,
       "step": 41400
     },
     {
       "epoch": 4.51,
-      "learning_rate": 5.4925600086890417e-05,
-      "loss": 2.8721,
+      "learning_rate": 5.492560008689042e-06,
+      "loss": 2.6017,
       "step": 41500
     },
     {
       "epoch": 4.52,
-      "learning_rate": 5.4816987075051596e-05,
-      "loss": 2.8518,
+      "learning_rate": 5.4816987075051595e-06,
+      "loss": 2.5957,
       "step": 41600
     },
     {
       "epoch": 4.53,
-      "learning_rate": 5.4708374063212776e-05,
-      "loss": 2.8708,
+      "learning_rate": 5.470837406321278e-06,
+      "loss": 2.6013,
       "step": 41700
     },
     {
       "epoch": 4.54,
-      "learning_rate": 5.4599761051373956e-05,
-      "loss": 2.8379,
+      "learning_rate": 5.459976105137396e-06,
+      "loss": 2.5761,
       "step": 41800
     },
     {
       "epoch": 4.55,
-      "learning_rate": 5.4491148039535136e-05,
-      "loss": 2.8438,
+      "learning_rate": 5.449114803953514e-06,
+      "loss": 2.5783,
       "step": 41900
     },
     {
       "epoch": 4.56,
-      "learning_rate": 5.438253502769632e-05,
-      "loss": 2.8823,
+      "learning_rate": 5.438253502769633e-06,
+      "loss": 2.6106,
       "step": 42000
     },
     {
       "epoch": 4.57,
-      "learning_rate": 5.42739220158575e-05,
-      "loss": 2.8721,
+      "learning_rate": 5.42739220158575e-06,
+      "loss": 2.6135,
       "step": 42100
     },
     {
       "epoch": 4.58,
-      "learning_rate": 5.416530900401868e-05,
-      "loss": 2.8483,
+      "learning_rate": 5.4165309004018684e-06,
+      "loss": 2.5915,
       "step": 42200
     },
     {
       "epoch": 4.59,
-      "learning_rate": 5.405669599217986e-05,
-      "loss": 2.828,
+      "learning_rate": 5.405669599217987e-06,
+      "loss": 2.5619,
       "step": 42300
     },
     {
       "epoch": 4.61,
-      "learning_rate": 5.394808298034105e-05,
-      "loss": 2.9338,
+      "learning_rate": 5.394808298034106e-06,
+      "loss": 2.6639,
       "step": 42400
     },
     {
       "epoch": 4.62,
-      "learning_rate": 5.383946996850223e-05,
-      "loss": 2.8982,
+      "learning_rate": 5.383946996850224e-06,
+      "loss": 2.6439,
       "step": 42500
     },
     {
       "epoch": 4.63,
-      "learning_rate": 5.373085695666341e-05,
-      "loss": 2.8346,
+      "learning_rate": 5.373085695666341e-06,
+      "loss": 2.5813,
       "step": 42600
     },
     {
       "epoch": 4.64,
-      "learning_rate": 5.362224394482459e-05,
-      "loss": 2.9424,
+      "learning_rate": 5.362224394482459e-06,
+      "loss": 2.6949,
       "step": 42700
     },
     {
       "epoch": 4.65,
-      "learning_rate": 5.351363093298577e-05,
-      "loss": 2.8952,
+      "learning_rate": 5.351363093298577e-06,
+      "loss": 2.6338,
       "step": 42800
     },
     {
       "epoch": 4.66,
-      "learning_rate": 5.340501792114696e-05,
-      "loss": 2.8737,
+      "learning_rate": 5.340501792114696e-06,
+      "loss": 2.6123,
       "step": 42900
     },
     {
       "epoch": 4.67,
-      "learning_rate": 5.329640490930814e-05,
-      "loss": 2.9484,
+      "learning_rate": 5.3296404909308146e-06,
+      "loss": 2.6859,
       "step": 43000
     },
     {
       "epoch": 4.68,
-      "learning_rate": 5.318779189746932e-05,
-      "loss": 2.8492,
+      "learning_rate": 5.318779189746932e-06,
+      "loss": 2.5906,
       "step": 43100
     },
     {
       "epoch": 4.69,
-      "learning_rate": 5.3079178885630495e-05,
-      "loss": 2.8864,
+      "learning_rate": 5.30791788856305e-06,
+      "loss": 2.6219,
       "step": 43200
     },
     {
       "epoch": 4.7,
-      "learning_rate": 5.2970565873791675e-05,
-      "loss": 2.8093,
+      "learning_rate": 5.297056587379168e-06,
+      "loss": 2.5597,
       "step": 43300
     },
     {
       "epoch": 4.71,
-      "learning_rate": 5.286195286195287e-05,
-      "loss": 2.7881,
+      "learning_rate": 5.286195286195287e-06,
+      "loss": 2.5275,
       "step": 43400
     },
     {
       "epoch": 4.72,
-      "learning_rate": 5.275333985011405e-05,
-      "loss": 2.8584,
+      "learning_rate": 5.2753339850114045e-06,
+      "loss": 2.6037,
       "step": 43500
     },
     {
       "epoch": 4.74,
-      "learning_rate": 5.264472683827523e-05,
-      "loss": 2.8925,
+      "learning_rate": 5.264472683827523e-06,
+      "loss": 2.627,
       "step": 43600
     },
     {
       "epoch": 4.75,
-      "learning_rate": 5.253611382643641e-05,
-      "loss": 2.7261,
+      "learning_rate": 5.253611382643641e-06,
+      "loss": 2.478,
       "step": 43700
     },
     {
       "epoch": 4.76,
-      "learning_rate": 5.2427500814597595e-05,
-      "loss": 2.8569,
+      "learning_rate": 5.24275008145976e-06,
+      "loss": 2.5955,
       "step": 43800
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.2318887802758775e-05,
-      "loss": 2.8978,
+      "learning_rate": 5.231888780275878e-06,
+      "loss": 2.6438,
       "step": 43900
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.2210274790919955e-05,
-      "loss": 2.9005,
+      "learning_rate": 5.221027479091995e-06,
+      "loss": 2.6428,
       "step": 44000
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.2101661779081135e-05,
-      "loss": 2.8661,
+      "learning_rate": 5.2101661779081135e-06,
+      "loss": 2.6154,
       "step": 44100
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.1993048767242315e-05,
-      "loss": 2.9499,
+      "learning_rate": 5.199304876724232e-06,
+      "loss": 2.6757,
       "step": 44200
     },
     {
       "epoch": 4.81,
-      "learning_rate": 5.18844357554035e-05,
-      "loss": 2.8969,
+      "learning_rate": 5.188443575540351e-06,
+      "loss": 2.655,
       "step": 44300
     },
     {
       "epoch": 4.82,
-      "learning_rate": 5.177582274356468e-05,
-      "loss": 2.8314,
+      "learning_rate": 5.177582274356469e-06,
+      "loss": 2.5796,
       "step": 44400
     },
     {
       "epoch": 4.83,
-      "learning_rate": 5.166720973172586e-05,
-      "loss": 2.8229,
+      "learning_rate": 5.166720973172586e-06,
+      "loss": 2.573,
       "step": 44500
     },
     {
       "epoch": 4.84,
-      "learning_rate": 5.155859671988704e-05,
-      "loss": 2.8463,
+      "learning_rate": 5.155859671988704e-06,
+      "loss": 2.5998,
       "step": 44600
     },
     {
       "epoch": 4.85,
-      "learning_rate": 5.144998370804822e-05,
-      "loss": 2.8764,
+      "learning_rate": 5.1449983708048224e-06,
+      "loss": 2.628,
       "step": 44700
     },
     {
       "epoch": 4.87,
-      "learning_rate": 5.1341370696209414e-05,
-      "loss": 2.8458,
+      "learning_rate": 5.1341370696209414e-06,
+      "loss": 2.6004,
       "step": 44800
     },
     {
       "epoch": 4.88,
-      "learning_rate": 5.1232757684370594e-05,
-      "loss": 2.8062,
+      "learning_rate": 5.12327576843706e-06,
+      "loss": 2.5632,
       "step": 44900
     },
     {
       "epoch": 4.89,
-      "learning_rate": 5.1124144672531774e-05,
-      "loss": 2.894,
+      "learning_rate": 5.112414467253177e-06,
+      "loss": 2.6409,
       "step": 45000
     },
     {
       "epoch": 4.9,
-      "learning_rate": 5.101553166069295e-05,
-      "loss": 2.8546,
+      "learning_rate": 5.101553166069295e-06,
+      "loss": 2.6091,
       "step": 45100
     },
     {
       "epoch": 4.91,
-      "learning_rate": 5.090691864885414e-05,
-      "loss": 2.8455,
+      "learning_rate": 5.090691864885414e-06,
+      "loss": 2.5956,
       "step": 45200
     },
     {
       "epoch": 4.92,
-      "learning_rate": 5.079830563701532e-05,
-      "loss": 2.8055,
+      "learning_rate": 5.079830563701532e-06,
+      "loss": 2.5589,
       "step": 45300
     },
     {
       "epoch": 4.93,
-      "learning_rate": 5.06896926251765e-05,
-      "loss": 2.9081,
+      "learning_rate": 5.06896926251765e-06,
+      "loss": 2.6596,
       "step": 45400
     },
     {
       "epoch": 4.94,
-      "learning_rate": 5.058107961333768e-05,
-      "loss": 2.9016,
+      "learning_rate": 5.058107961333768e-06,
+      "loss": 2.6573,
       "step": 45500
     },
     {
       "epoch": 4.95,
-      "learning_rate": 5.047246660149886e-05,
-      "loss": 2.88,
+      "learning_rate": 5.047246660149886e-06,
+      "loss": 2.6297,
       "step": 45600
     },
     {
       "epoch": 4.96,
-      "learning_rate": 5.036385358966005e-05,
-      "loss": 2.9696,
+      "learning_rate": 5.036385358966005e-06,
+      "loss": 2.7212,
       "step": 45700
     },
     {
       "epoch": 4.97,
-      "learning_rate": 5.025524057782123e-05,
-      "loss": 2.8584,
+      "learning_rate": 5.025524057782123e-06,
+      "loss": 2.6198,
       "step": 45800
     },
     {
       "epoch": 4.99,
-      "learning_rate": 5.014662756598241e-05,
-      "loss": 2.8707,
+      "learning_rate": 5.014662756598241e-06,
+      "loss": 2.637,
       "step": 45900
     },
     {
       "epoch": 5.0,
-      "learning_rate": 5.003801455414359e-05,
-      "loss": 2.8539,
+      "learning_rate": 5.0038014554143585e-06,
+      "loss": 2.6052,
       "step": 46000
     },
     {
       "epoch": 5.01,
-      "learning_rate": 4.9929401542304773e-05,
-      "loss": 2.8842,
+      "learning_rate": 4.9929401542304775e-06,
+      "loss": 2.6623,
       "step": 46100
     },
     {
       "epoch": 5.02,
-      "learning_rate": 4.982078853046595e-05,
-      "loss": 2.7645,
+      "learning_rate": 4.982078853046595e-06,
+      "loss": 2.5413,
       "step": 46200
     },
     {
       "epoch": 5.03,
-      "learning_rate": 4.971217551862713e-05,
-      "loss": 2.8499,
+      "learning_rate": 4.971217551862714e-06,
+      "loss": 2.6243,
       "step": 46300
     },
     {
       "epoch": 5.04,
-      "learning_rate": 4.960356250678831e-05,
-      "loss": 2.8812,
+      "learning_rate": 4.960356250678832e-06,
+      "loss": 2.6549,
       "step": 46400
     },
     {
       "epoch": 5.05,
-      "learning_rate": 4.94949494949495e-05,
-      "loss": 2.7704,
+      "learning_rate": 4.94949494949495e-06,
+      "loss": 2.5527,
       "step": 46500
     },
     {
       "epoch": 5.06,
-      "learning_rate": 4.938633648311068e-05,
-      "loss": 2.7921,
+      "learning_rate": 4.938633648311068e-06,
+      "loss": 2.5732,
       "step": 46600
     },
     {
       "epoch": 5.07,
-      "learning_rate": 4.927772347127186e-05,
-      "loss": 2.8428,
+      "learning_rate": 4.927772347127186e-06,
+      "loss": 2.6203,
       "step": 46700
     },
     {
       "epoch": 5.08,
-      "learning_rate": 4.9169110459433046e-05,
-      "loss": 2.8605,
+      "learning_rate": 4.916911045943305e-06,
+      "loss": 2.6308,
       "step": 46800
     },
     {
       "epoch": 5.09,
-      "learning_rate": 4.906049744759422e-05,
-      "loss": 2.7879,
+      "learning_rate": 4.906049744759423e-06,
+      "loss": 2.5752,
       "step": 46900
     },
     {
       "epoch": 5.1,
-      "learning_rate": 4.8951884435755406e-05,
-      "loss": 2.846,
+      "learning_rate": 4.895188443575541e-06,
+      "loss": 2.6185,
       "step": 47000
     },
     {
       "epoch": 5.12,
-      "learning_rate": 4.8843271423916586e-05,
-      "loss": 2.8179,
+      "learning_rate": 4.884327142391659e-06,
+      "loss": 2.5957,
       "step": 47100
     },
     {
       "epoch": 5.13,
-      "learning_rate": 4.873465841207777e-05,
-      "loss": 2.9302,
+      "learning_rate": 4.873465841207777e-06,
+      "loss": 2.7052,
       "step": 47200
     },
     {
       "epoch": 5.14,
-      "learning_rate": 4.862604540023895e-05,
-      "loss": 2.7913,
+      "learning_rate": 4.8626045400238954e-06,
+      "loss": 2.5755,
       "step": 47300
     },
     {
       "epoch": 5.15,
-      "learning_rate": 4.851743238840013e-05,
-      "loss": 2.7173,
+      "learning_rate": 4.851743238840014e-06,
+      "loss": 2.5051,
       "step": 47400
     },
     {
       "epoch": 5.16,
-      "learning_rate": 4.840881937656131e-05,
-      "loss": 2.8389,
+      "learning_rate": 4.840881937656132e-06,
+      "loss": 2.6183,
       "step": 47500
     },
     {
       "epoch": 5.17,
-      "learning_rate": 4.830020636472249e-05,
-      "loss": 2.6752,
+      "learning_rate": 4.83002063647225e-06,
+      "loss": 2.4524,
       "step": 47600
     },
     {
       "epoch": 5.18,
-      "learning_rate": 4.819159335288368e-05,
-      "loss": 2.7911,
+      "learning_rate": 4.819159335288368e-06,
+      "loss": 2.578,
       "step": 47700
     },
     {
       "epoch": 5.19,
-      "learning_rate": 4.808298034104486e-05,
-      "loss": 2.8147,
+      "learning_rate": 4.808298034104486e-06,
+      "loss": 2.598,
       "step": 47800
     },
     {
       "epoch": 5.2,
-      "learning_rate": 4.7974367329206046e-05,
-      "loss": 2.7192,
+      "learning_rate": 4.797436732920604e-06,
+      "loss": 2.5113,
       "step": 47900
     },
     {
       "epoch": 5.21,
-      "learning_rate": 4.7865754317367225e-05,
-      "loss": 2.8325,
+      "learning_rate": 4.7865754317367225e-06,
+      "loss": 2.6115,
       "step": 48000
     },
     {
       "epoch": 5.22,
-      "learning_rate": 4.7757141305528405e-05,
-      "loss": 2.8132,
+      "learning_rate": 4.775714130552841e-06,
+      "loss": 2.5991,
       "step": 48100
     },
     {
       "epoch": 5.24,
-      "learning_rate": 4.7648528293689585e-05,
-      "loss": 2.7436,
+      "learning_rate": 4.764852829368959e-06,
+      "loss": 2.5349,
       "step": 48200
     },
     {
       "epoch": 5.25,
-      "learning_rate": 4.7539915281850765e-05,
-      "loss": 2.7892,
+      "learning_rate": 4.753991528185077e-06,
+      "loss": 2.5741,
       "step": 48300
     },
     {
       "epoch": 5.26,
-      "learning_rate": 4.743130227001195e-05,
-      "loss": 2.7921,
+      "learning_rate": 4.743130227001195e-06,
+      "loss": 2.571,
       "step": 48400
     },
     {
       "epoch": 5.27,
-      "learning_rate": 4.732268925817313e-05,
-      "loss": 2.7845,
+      "learning_rate": 4.732268925817313e-06,
+      "loss": 2.5685,
       "step": 48500
     },
     {
       "epoch": 5.28,
-      "learning_rate": 4.721407624633432e-05,
-      "loss": 2.8523,
+      "learning_rate": 4.7214076246334315e-06,
+      "loss": 2.6359,
       "step": 48600
     },
     {
       "epoch": 5.29,
-      "learning_rate": 4.71054632344955e-05,
-      "loss": 2.8484,
+      "learning_rate": 4.71054632344955e-06,
+      "loss": 2.6265,
       "step": 48700
     },
     {
       "epoch": 5.3,
-      "learning_rate": 4.699685022265667e-05,
-      "loss": 2.7924,
+      "learning_rate": 4.699685022265668e-06,
+      "loss": 2.584,
       "step": 48800
     },
     {
       "epoch": 5.31,
-      "learning_rate": 4.688823721081786e-05,
-      "loss": 2.8363,
+      "learning_rate": 4.688823721081786e-06,
+      "loss": 2.624,
       "step": 48900
     },
     {
       "epoch": 5.32,
-      "learning_rate": 4.677962419897904e-05,
-      "loss": 2.7945,
+      "learning_rate": 4.677962419897904e-06,
+      "loss": 2.5784,
       "step": 49000
     },
     {
       "epoch": 5.33,
-      "learning_rate": 4.6671011187140225e-05,
-      "loss": 2.8128,
+      "learning_rate": 4.667101118714022e-06,
+      "loss": 2.5947,
       "step": 49100
     },
     {
       "epoch": 5.34,
-      "learning_rate": 4.6562398175301405e-05,
-      "loss": 2.8558,
+      "learning_rate": 4.6562398175301405e-06,
+      "loss": 2.6441,
       "step": 49200
     },
     {
       "epoch": 5.35,
-      "learning_rate": 4.6453785163462584e-05,
-      "loss": 2.838,
+      "learning_rate": 4.645378516346259e-06,
+      "loss": 2.6232,
       "step": 49300
     },
     {
       "epoch": 5.37,
-      "learning_rate": 4.6345172151623764e-05,
-      "loss": 2.7965,
+      "learning_rate": 4.634517215162377e-06,
+      "loss": 2.581,
       "step": 49400
     },
     {
       "epoch": 5.38,
-      "learning_rate": 4.6236559139784944e-05,
-      "loss": 2.742,
+      "learning_rate": 4.623655913978495e-06,
+      "loss": 2.5392,
       "step": 49500
     },
     {
       "epoch": 5.39,
-      "learning_rate": 4.612794612794613e-05,
-      "loss": 2.6993,
+      "learning_rate": 4.612794612794613e-06,
+      "loss": 2.5009,
       "step": 49600
     },
     {
       "epoch": 5.4,
-      "learning_rate": 4.601933311610731e-05,
-      "loss": 2.8197,
+      "learning_rate": 4.601933311610731e-06,
+      "loss": 2.6065,
       "step": 49700
     },
     {
       "epoch": 5.41,
-      "learning_rate": 4.59107201042685e-05,
-      "loss": 2.7072,
+      "learning_rate": 4.591072010426849e-06,
+      "loss": 2.5019,
       "step": 49800
     },
     {
       "epoch": 5.42,
-      "learning_rate": 4.580210709242968e-05,
-      "loss": 2.8693,
+      "learning_rate": 4.580210709242968e-06,
+      "loss": 2.6518,
       "step": 49900
     },
     {
       "epoch": 5.43,
-      "learning_rate": 4.569349408059086e-05,
-      "loss": 2.795,
+      "learning_rate": 4.569349408059086e-06,
+      "loss": 2.5883,
       "step": 50000
     },
     {
       "epoch": 5.44,
-      "learning_rate": 4.558488106875204e-05,
-      "loss": 2.8191,
+      "learning_rate": 4.558488106875204e-06,
+      "loss": 2.6121,
       "step": 50100
     },
     {
       "epoch": 5.45,
-      "learning_rate": 4.547626805691322e-05,
-      "loss": 2.8621,
+      "learning_rate": 4.547626805691322e-06,
+      "loss": 2.6546,
       "step": 50200
     },
     {
       "epoch": 5.46,
-      "learning_rate": 4.5367655045074404e-05,
-      "loss": 2.7651,
+      "learning_rate": 4.53676550450744e-06,
+      "loss": 2.5675,
       "step": 50300
     },
     {
       "epoch": 5.47,
-      "learning_rate": 4.5259042033235584e-05,
-      "loss": 2.77,
+      "learning_rate": 4.525904203323558e-06,
+      "loss": 2.5605,
       "step": 50400
     },
     {
       "epoch": 5.48,
-      "learning_rate": 4.515042902139677e-05,
-      "loss": 2.8439,
+      "learning_rate": 4.5150429021396765e-06,
+      "loss": 2.6376,
       "step": 50500
     },
     {
       "epoch": 5.5,
-      "learning_rate": 4.5041816009557944e-05,
-      "loss": 2.7788,
+      "learning_rate": 4.504181600955795e-06,
+      "loss": 2.573,
       "step": 50600
     },
     {
       "epoch": 5.51,
-      "learning_rate": 4.493320299771913e-05,
-      "loss": 2.8328,
+      "learning_rate": 4.493320299771914e-06,
+      "loss": 2.625,
       "step": 50700
     },
     {
       "epoch": 5.52,
-      "learning_rate": 4.482458998588031e-05,
-      "loss": 2.8625,
+      "learning_rate": 4.482458998588031e-06,
+      "loss": 2.6583,
       "step": 50800
     },
     {
       "epoch": 5.53,
-      "learning_rate": 4.471597697404149e-05,
-      "loss": 2.8773,
+      "learning_rate": 4.471597697404149e-06,
+      "loss": 2.6753,
       "step": 50900
     },
     {
       "epoch": 5.54,
-      "learning_rate": 4.460736396220268e-05,
-      "loss": 2.8209,
+      "learning_rate": 4.460736396220267e-06,
+      "loss": 2.6153,
       "step": 51000
     },
     {
       "epoch": 5.55,
-      "learning_rate": 4.449875095036386e-05,
-      "loss": 2.7392,
+      "learning_rate": 4.4498750950363855e-06,
+      "loss": 2.5356,
       "step": 51100
     },
     {
       "epoch": 5.56,
-      "learning_rate": 4.4390137938525037e-05,
-      "loss": 2.7048,
+      "learning_rate": 4.4390137938525045e-06,
+      "loss": 2.5011,
       "step": 51200
     },
     {
       "epoch": 5.57,
-      "learning_rate": 4.4281524926686216e-05,
-      "loss": 2.8771,
+      "learning_rate": 4.428152492668622e-06,
+      "loss": 2.6799,
       "step": 51300
     },
     {
       "epoch": 5.58,
-      "learning_rate": 4.41729119148474e-05,
-      "loss": 2.7526,
+      "learning_rate": 4.417291191484741e-06,
+      "loss": 2.5622,
       "step": 51400
     },
     {
       "epoch": 5.59,
-      "learning_rate": 4.406429890300858e-05,
-      "loss": 2.7457,
+      "learning_rate": 4.406429890300858e-06,
+      "loss": 2.5409,
       "step": 51500
     },
     {
       "epoch": 5.6,
-      "learning_rate": 4.395568589116976e-05,
-      "loss": 2.8338,
+      "learning_rate": 4.395568589116976e-06,
+      "loss": 2.6347,
       "step": 51600
     },
     {
       "epoch": 5.62,
-      "learning_rate": 4.384707287933095e-05,
-      "loss": 2.8154,
+      "learning_rate": 4.3847072879330945e-06,
+      "loss": 2.6162,
       "step": 51700
     },
     {
       "epoch": 5.63,
-      "learning_rate": 4.373845986749213e-05,
-      "loss": 2.7648,
+      "learning_rate": 4.373845986749213e-06,
+      "loss": 2.5708,
       "step": 51800
     },
     {
       "epoch": 5.64,
-      "learning_rate": 4.362984685565331e-05,
-      "loss": 2.81,
+      "learning_rate": 4.362984685565332e-06,
+      "loss": 2.6125,
       "step": 51900
     },
     {
       "epoch": 5.65,
-      "learning_rate": 4.352123384381449e-05,
-      "loss": 2.742,
+      "learning_rate": 4.352123384381449e-06,
+      "loss": 2.5505,
       "step": 52000
     },
     {
       "epoch": 5.66,
-      "learning_rate": 4.3412620831975676e-05,
-      "loss": 2.7741,
+      "learning_rate": 4.341262083197568e-06,
+      "loss": 2.5717,
       "step": 52100
     },
     {
       "epoch": 5.67,
-      "learning_rate": 4.3304007820136856e-05,
-      "loss": 2.7733,
+      "learning_rate": 4.330400782013685e-06,
+      "loss": 2.5799,
       "step": 52200
     },
     {
       "epoch": 5.68,
-      "learning_rate": 4.3195394808298036e-05,
-      "loss": 2.8178,
+      "learning_rate": 4.319539480829803e-06,
+      "loss": 2.6216,
       "step": 52300
     },
     {
       "epoch": 5.69,
-      "learning_rate": 4.308678179645922e-05,
-      "loss": 2.8433,
+      "learning_rate": 4.308678179645922e-06,
+      "loss": 2.6444,
       "step": 52400
     },
     {
       "epoch": 5.7,
-      "learning_rate": 4.2978168784620396e-05,
-      "loss": 2.8015,
+      "learning_rate": 4.29781687846204e-06,
+      "loss": 2.6076,
       "step": 52500
     },
     {
       "epoch": 5.71,
-      "learning_rate": 4.286955577278158e-05,
-      "loss": 2.8093,
+      "learning_rate": 4.286955577278159e-06,
+      "loss": 2.6124,
       "step": 52600
     },
     {
       "epoch": 5.72,
-      "learning_rate": 4.276094276094276e-05,
-      "loss": 2.8235,
+      "learning_rate": 4.276094276094276e-06,
+      "loss": 2.6234,
       "step": 52700
     },
     {
       "epoch": 5.73,
-      "learning_rate": 4.265232974910394e-05,
-      "loss": 2.7772,
+      "learning_rate": 4.265232974910394e-06,
+      "loss": 2.5742,
       "step": 52800
     },
     {
       "epoch": 5.75,
-      "learning_rate": 4.254371673726513e-05,
-      "loss": 2.7632,
+      "learning_rate": 4.254371673726513e-06,
+      "loss": 2.5713,
       "step": 52900
     },
     {
       "epoch": 5.76,
-      "learning_rate": 4.243510372542631e-05,
-      "loss": 2.9062,
+      "learning_rate": 4.2435103725426305e-06,
+      "loss": 2.713,
       "step": 53000
     },
     {
       "epoch": 5.77,
-      "learning_rate": 4.232649071358749e-05,
-      "loss": 2.7796,
+      "learning_rate": 4.2326490713587495e-06,
+      "loss": 2.5845,
       "step": 53100
     },
     {
       "epoch": 5.78,
-      "learning_rate": 4.221787770174867e-05,
-      "loss": 2.7974,
+      "learning_rate": 4.221787770174867e-06,
+      "loss": 2.6055,
       "step": 53200
     },
     {
       "epoch": 5.79,
-      "learning_rate": 4.2109264689909855e-05,
-      "loss": 2.8285,
+      "learning_rate": 4.210926468990986e-06,
+      "loss": 2.6323,
       "step": 53300
     },
     {
       "epoch": 5.8,
-      "learning_rate": 4.2000651678071035e-05,
-      "loss": 2.8202,
+      "learning_rate": 4.200065167807104e-06,
+      "loss": 2.624,
       "step": 53400
     },
     {
       "epoch": 5.81,
-      "learning_rate": 4.1892038666232215e-05,
-      "loss": 2.8093,
+      "learning_rate": 4.189203866623221e-06,
+      "loss": 2.6161,
       "step": 53500
     },
     {
       "epoch": 5.82,
-      "learning_rate": 4.17834256543934e-05,
-      "loss": 2.8685,
+      "learning_rate": 4.17834256543934e-06,
+      "loss": 2.6798,
       "step": 53600
     },
     {
       "epoch": 5.83,
-      "learning_rate": 4.167481264255458e-05,
-      "loss": 2.8321,
+      "learning_rate": 4.167481264255458e-06,
+      "loss": 2.6378,
       "step": 53700
     },
     {
       "epoch": 5.84,
-      "learning_rate": 4.156619963071576e-05,
-      "loss": 2.6883,
+      "learning_rate": 4.156619963071577e-06,
+      "loss": 2.4997,
       "step": 53800
     },
     {
       "epoch": 5.85,
-      "learning_rate": 4.145758661887694e-05,
-      "loss": 2.7624,
+      "learning_rate": 4.145758661887694e-06,
+      "loss": 2.5783,
       "step": 53900
     },
     {
       "epoch": 5.87,
-      "learning_rate": 4.134897360703813e-05,
-      "loss": 2.8609,
+      "learning_rate": 4.134897360703813e-06,
+      "loss": 2.6712,
       "step": 54000
     },
     {
       "epoch": 5.88,
-      "learning_rate": 4.124036059519931e-05,
-      "loss": 2.8817,
+      "learning_rate": 4.124036059519931e-06,
+      "loss": 2.6916,
       "step": 54100
     },
     {
       "epoch": 5.89,
-      "learning_rate": 4.113174758336049e-05,
-      "loss": 2.8336,
+      "learning_rate": 4.1131747583360484e-06,
+      "loss": 2.6379,
       "step": 54200
     },
     {
       "epoch": 5.9,
-      "learning_rate": 4.102313457152167e-05,
-      "loss": 2.8221,
+      "learning_rate": 4.1023134571521674e-06,
+      "loss": 2.6296,
       "step": 54300
     },
     {
       "epoch": 5.91,
-      "learning_rate": 4.091452155968285e-05,
-      "loss": 2.7564,
+      "learning_rate": 4.091452155968285e-06,
+      "loss": 2.5687,
       "step": 54400
     },
     {
       "epoch": 5.92,
-      "learning_rate": 4.0805908547844034e-05,
-      "loss": 2.796,
+      "learning_rate": 4.080590854784404e-06,
+      "loss": 2.6019,
       "step": 54500
     },
     {
       "epoch": 5.93,
-      "learning_rate": 4.0697295536005214e-05,
-      "loss": 2.7683,
+      "learning_rate": 4.069729553600522e-06,
+      "loss": 2.5791,
       "step": 54600
     },
     {
       "epoch": 5.94,
-      "learning_rate": 4.05886825241664e-05,
-      "loss": 2.711,
+      "learning_rate": 4.05886825241664e-06,
+      "loss": 2.5322,
       "step": 54700
     },
     {
       "epoch": 5.95,
-      "learning_rate": 4.048006951232758e-05,
-      "loss": 2.7996,
+      "learning_rate": 4.048006951232758e-06,
+      "loss": 2.6095,
       "step": 54800
     },
     {
       "epoch": 5.96,
-      "learning_rate": 4.037145650048876e-05,
-      "loss": 2.7824,
+      "learning_rate": 4.0371456500488756e-06,
+      "loss": 2.5959,
       "step": 54900
     },
     {
       "epoch": 5.97,
-      "learning_rate": 4.026284348864994e-05,
-      "loss": 2.8153,
+      "learning_rate": 4.0262843488649946e-06,
+      "loss": 2.6322,
       "step": 55000
     },
     {
       "epoch": 5.98,
-      "learning_rate": 4.015423047681112e-05,
-      "loss": 2.8316,
+      "learning_rate": 4.015423047681113e-06,
+      "loss": 2.6329,
       "step": 55100
     },
     {
       "epoch": 6.0,
-      "learning_rate": 4.004561746497231e-05,
-      "loss": 2.7809,
+      "learning_rate": 4.004561746497231e-06,
+      "loss": 2.5924,
       "step": 55200
     },
     {
       "epoch": 6.01,
-      "learning_rate": 3.993700445313349e-05,
-      "loss": 2.745,
+      "learning_rate": 3.993700445313349e-06,
+      "loss": 2.5712,
       "step": 55300
     },
     {
       "epoch": 6.02,
-      "learning_rate": 3.9828391441294674e-05,
-      "loss": 2.7401,
+      "learning_rate": 3.982839144129467e-06,
+      "loss": 2.5713,
       "step": 55400
     },
     {
       "epoch": 6.03,
-      "learning_rate": 3.9719778429455854e-05,
-      "loss": 2.7371,
+      "learning_rate": 3.971977842945585e-06,
+      "loss": 2.5672,
       "step": 55500
     },
     {
       "epoch": 6.04,
-      "learning_rate": 3.961116541761703e-05,
-      "loss": 2.723,
+      "learning_rate": 3.9611165417617035e-06,
+      "loss": 2.5538,
       "step": 55600
     },
     {
       "epoch": 6.05,
-      "learning_rate": 3.9502552405778213e-05,
-      "loss": 2.7057,
+      "learning_rate": 3.950255240577822e-06,
+      "loss": 2.5359,
       "step": 55700
     },
     {
       "epoch": 6.06,
-      "learning_rate": 3.939393939393939e-05,
-      "loss": 2.8222,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 2.6465,
       "step": 55800
     },
     {
       "epoch": 6.07,
-      "learning_rate": 3.928532638210058e-05,
-      "loss": 2.6662,
+      "learning_rate": 3.928532638210058e-06,
+      "loss": 2.4974,
       "step": 55900
     },
     {
       "epoch": 6.08,
-      "learning_rate": 3.917671337026176e-05,
-      "loss": 2.7647,
+      "learning_rate": 3.917671337026176e-06,
+      "loss": 2.5944,
       "step": 56000
     },
     {
       "epoch": 6.09,
-      "learning_rate": 3.906810035842295e-05,
-      "loss": 2.7936,
+      "learning_rate": 3.906810035842294e-06,
+      "loss": 2.627,
       "step": 56100
     },
     {
       "epoch": 6.1,
-      "learning_rate": 3.895948734658412e-05,
-      "loss": 2.7848,
+      "learning_rate": 3.8959487346584125e-06,
+      "loss": 2.6114,
       "step": 56200
     },
     {
       "epoch": 6.11,
-      "learning_rate": 3.88508743347453e-05,
-      "loss": 2.768,
+      "learning_rate": 3.885087433474531e-06,
+      "loss": 2.5997,
       "step": 56300
     },
     {
       "epoch": 6.13,
-      "learning_rate": 3.8742261322906486e-05,
-      "loss": 2.7741,
+      "learning_rate": 3.874226132290649e-06,
+      "loss": 2.5989,
       "step": 56400
     },
     {
       "epoch": 6.14,
-      "learning_rate": 3.8633648311067666e-05,
-      "loss": 2.7476,
+      "learning_rate": 3.863364831106767e-06,
+      "loss": 2.58,
       "step": 56500
     },
     {
       "epoch": 6.15,
-      "learning_rate": 3.852503529922885e-05,
-      "loss": 2.778,
+      "learning_rate": 3.852503529922885e-06,
+      "loss": 2.6062,
       "step": 56600
     },
     {
       "epoch": 6.16,
-      "learning_rate": 3.841642228739003e-05,
-      "loss": 2.8299,
+      "learning_rate": 3.841642228739003e-06,
+      "loss": 2.6589,
       "step": 56700
     },
     {
       "epoch": 6.17,
-      "learning_rate": 3.830780927555121e-05,
-      "loss": 2.785,
+      "learning_rate": 3.8307809275551214e-06,
+      "loss": 2.6167,
       "step": 56800
     },
     {
       "epoch": 6.18,
-      "learning_rate": 3.819919626371239e-05,
-      "loss": 2.6849,
+      "learning_rate": 3.81991962637124e-06,
+      "loss": 2.5203,
       "step": 56900
     },
     {
       "epoch": 6.19,
-      "learning_rate": 3.809058325187357e-05,
-      "loss": 2.6925,
+      "learning_rate": 3.8090583251873573e-06,
+      "loss": 2.5281,
       "step": 57000
     },
     {
       "epoch": 6.2,
-      "learning_rate": 3.798197024003476e-05,
-      "loss": 2.6585,
+      "learning_rate": 3.798197024003476e-06,
+      "loss": 2.5002,
       "step": 57100
     },
     {
       "epoch": 6.21,
-      "learning_rate": 3.787335722819594e-05,
-      "loss": 2.7174,
+      "learning_rate": 3.787335722819594e-06,
+      "loss": 2.5477,
       "step": 57200
     },
     {
       "epoch": 6.22,
-      "learning_rate": 3.7764744216357126e-05,
-      "loss": 2.8318,
+      "learning_rate": 3.7764744216357122e-06,
+      "loss": 2.667,
       "step": 57300
     },
     {
       "epoch": 6.23,
-      "learning_rate": 3.7656131204518306e-05,
-      "loss": 2.7448,
+      "learning_rate": 3.7656131204518304e-06,
+      "loss": 2.5773,
       "step": 57400
     },
     {
       "epoch": 6.25,
-      "learning_rate": 3.7547518192679486e-05,
-      "loss": 2.8137,
+      "learning_rate": 3.754751819267949e-06,
+      "loss": 2.6551,
       "step": 57500
     },
     {
       "epoch": 6.26,
-      "learning_rate": 3.7438905180840665e-05,
-      "loss": 2.6344,
+      "learning_rate": 3.7438905180840667e-06,
+      "loss": 2.4766,
       "step": 57600
     },
     {
       "epoch": 6.27,
-      "learning_rate": 3.7330292169001845e-05,
-      "loss": 2.7126,
+      "learning_rate": 3.733029216900185e-06,
+      "loss": 2.55,
       "step": 57700
     },
     {
       "epoch": 6.28,
-      "learning_rate": 3.722167915716303e-05,
-      "loss": 2.7564,
+      "learning_rate": 3.722167915716303e-06,
+      "loss": 2.5876,
       "step": 57800
     },
     {
       "epoch": 6.29,
-      "learning_rate": 3.711306614532421e-05,
-      "loss": 2.7657,
+      "learning_rate": 3.711306614532421e-06,
+      "loss": 2.6051,
       "step": 57900
     },
     {
       "epoch": 6.3,
-      "learning_rate": 3.700445313348539e-05,
-      "loss": 2.8002,
+      "learning_rate": 3.7004453133485398e-06,
+      "loss": 2.6491,
       "step": 58000
     },
     {
       "epoch": 6.31,
-      "learning_rate": 3.689584012164657e-05,
-      "loss": 2.7179,
+      "learning_rate": 3.6895840121646575e-06,
+      "loss": 2.5531,
       "step": 58100
     },
     {
       "epoch": 6.32,
-      "learning_rate": 3.678722710980776e-05,
-      "loss": 2.7635,
+      "learning_rate": 3.678722710980776e-06,
+      "loss": 2.599,
       "step": 58200
     },
     {
       "epoch": 6.33,
-      "learning_rate": 3.667861409796894e-05,
-      "loss": 2.6482,
+      "learning_rate": 3.667861409796894e-06,
+      "loss": 2.4885,
       "step": 58300
     },
     {
       "epoch": 6.34,
-      "learning_rate": 3.657000108613012e-05,
-      "loss": 2.7555,
+      "learning_rate": 3.657000108613012e-06,
+      "loss": 2.5962,
       "step": 58400
     },
     {
       "epoch": 6.35,
-      "learning_rate": 3.6461388074291305e-05,
-      "loss": 2.7954,
+      "learning_rate": 3.6461388074291306e-06,
+      "loss": 2.6307,
       "step": 58500
     },
     {
       "epoch": 6.36,
-      "learning_rate": 3.6352775062452485e-05,
-      "loss": 2.7937,
+      "learning_rate": 3.6352775062452483e-06,
+      "loss": 2.6305,
       "step": 58600
     },
     {
       "epoch": 6.38,
-      "learning_rate": 3.6244162050613665e-05,
-      "loss": 2.7426,
+      "learning_rate": 3.624416205061367e-06,
+      "loss": 2.5787,
       "step": 58700
     },
     {
       "epoch": 6.39,
-      "learning_rate": 3.6135549038774845e-05,
-      "loss": 2.8019,
+      "learning_rate": 3.6135549038774846e-06,
+      "loss": 2.6416,
       "step": 58800
     },
     {
       "epoch": 6.4,
-      "learning_rate": 3.602693602693603e-05,
-      "loss": 2.7112,
+      "learning_rate": 3.6026936026936032e-06,
+      "loss": 2.5507,
       "step": 58900
     },
     {
       "epoch": 6.41,
-      "learning_rate": 3.591832301509721e-05,
-      "loss": 2.809,
+      "learning_rate": 3.5918323015097214e-06,
+      "loss": 2.6438,
       "step": 59000
     },
     {
       "epoch": 6.42,
-      "learning_rate": 3.580971000325839e-05,
-      "loss": 2.7244,
+      "learning_rate": 3.580971000325839e-06,
+      "loss": 2.5636,
       "step": 59100
     },
     {
       "epoch": 6.43,
-      "learning_rate": 3.570109699141958e-05,
-      "loss": 2.7719,
+      "learning_rate": 3.5701096991419577e-06,
+      "loss": 2.6248,
       "step": 59200
     },
     {
       "epoch": 6.44,
-      "learning_rate": 3.559248397958075e-05,
-      "loss": 2.9352,
+      "learning_rate": 3.5592483979580754e-06,
+      "loss": 2.7704,
       "step": 59300
     },
     {
       "epoch": 6.45,
-      "learning_rate": 3.548387096774194e-05,
-      "loss": 2.711,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 2.5512,
       "step": 59400
     },
     {
       "epoch": 6.46,
-      "learning_rate": 3.537525795590312e-05,
-      "loss": 2.7434,
+      "learning_rate": 3.5375257955903118e-06,
+      "loss": 2.5806,
       "step": 59500
     },
     {
       "epoch": 6.47,
-      "learning_rate": 3.5266644944064304e-05,
-      "loss": 2.7573,
+      "learning_rate": 3.5266644944064303e-06,
+      "loss": 2.6054,
       "step": 59600
     },
     {
       "epoch": 6.48,
-      "learning_rate": 3.5158031932225484e-05,
-      "loss": 2.6963,
+      "learning_rate": 3.5158031932225485e-06,
+      "loss": 2.5407,
       "step": 59700
     },
     {
       "epoch": 6.5,
-      "learning_rate": 3.5049418920386664e-05,
-      "loss": 2.8267,
+      "learning_rate": 3.5049418920386662e-06,
+      "loss": 2.6659,
       "step": 59800
     },
     {
       "epoch": 6.51,
-      "learning_rate": 3.4940805908547844e-05,
-      "loss": 2.7864,
+      "learning_rate": 3.494080590854785e-06,
+      "loss": 2.6348,
       "step": 59900
     },
     {
       "epoch": 6.52,
-      "learning_rate": 3.4832192896709024e-05,
-      "loss": 2.7585,
+      "learning_rate": 3.4832192896709026e-06,
+      "loss": 2.6008,
       "step": 60000
     },
     {
       "epoch": 6.53,
-      "learning_rate": 3.472357988487021e-05,
-      "loss": 2.714,
+      "learning_rate": 3.472357988487021e-06,
+      "loss": 2.5594,
       "step": 60100
     },
     {
       "epoch": 6.54,
-      "learning_rate": 3.461496687303139e-05,
-      "loss": 2.7202,
+      "learning_rate": 3.4614966873031393e-06,
+      "loss": 2.5685,
       "step": 60200
     },
     {
       "epoch": 6.55,
-      "learning_rate": 3.450635386119258e-05,
-      "loss": 2.8636,
+      "learning_rate": 3.4506353861192575e-06,
+      "loss": 2.7101,
       "step": 60300
     },
     {
       "epoch": 6.56,
-      "learning_rate": 3.439774084935376e-05,
-      "loss": 2.8138,
+      "learning_rate": 3.4397740849353756e-06,
+      "loss": 2.6515,
       "step": 60400
     },
     {
       "epoch": 6.57,
-      "learning_rate": 3.428912783751494e-05,
-      "loss": 2.74,
+      "learning_rate": 3.4289127837514933e-06,
+      "loss": 2.5911,
       "step": 60500
     },
     {
       "epoch": 6.58,
-      "learning_rate": 3.418051482567612e-05,
-      "loss": 2.7704,
+      "learning_rate": 3.418051482567612e-06,
+      "loss": 2.6185,
       "step": 60600
     },
     {
       "epoch": 6.59,
-      "learning_rate": 3.40719018138373e-05,
-      "loss": 2.7326,
+      "learning_rate": 3.40719018138373e-06,
+      "loss": 2.5763,
       "step": 60700
     },
     {
       "epoch": 6.6,
-      "learning_rate": 3.396328880199848e-05,
-      "loss": 2.7531,
+      "learning_rate": 3.3963288801998483e-06,
+      "loss": 2.6032,
       "step": 60800
     },
     {
       "epoch": 6.61,
-      "learning_rate": 3.385467579015966e-05,
-      "loss": 2.67,
+      "learning_rate": 3.3854675790159664e-06,
+      "loss": 2.5147,
       "step": 60900
     },
     {
       "epoch": 6.63,
-      "learning_rate": 3.374606277832085e-05,
-      "loss": 2.7537,
+      "learning_rate": 3.374606277832085e-06,
+      "loss": 2.6022,
       "step": 61000
     },
     {
       "epoch": 6.64,
-      "learning_rate": 3.363744976648203e-05,
-      "loss": 2.7866,
+      "learning_rate": 3.3637449766482027e-06,
+      "loss": 2.6284,
       "step": 61100
     },
     {
       "epoch": 6.65,
-      "learning_rate": 3.35288367546432e-05,
-      "loss": 2.7174,
+      "learning_rate": 3.352883675464321e-06,
+      "loss": 2.5661,
       "step": 61200
     },
     {
       "epoch": 6.66,
-      "learning_rate": 3.342022374280439e-05,
-      "loss": 2.7214,
+      "learning_rate": 3.342022374280439e-06,
+      "loss": 2.5742,
       "step": 61300
     },
     {
       "epoch": 6.67,
-      "learning_rate": 3.331161073096557e-05,
-      "loss": 2.7205,
+      "learning_rate": 3.331161073096557e-06,
+      "loss": 2.566,
       "step": 61400
     },
     {
       "epoch": 6.68,
-      "learning_rate": 3.3202997719126756e-05,
-      "loss": 2.8244,
+      "learning_rate": 3.320299771912676e-06,
+      "loss": 2.6662,
       "step": 61500
     },
     {
       "epoch": 6.69,
-      "learning_rate": 3.3094384707287936e-05,
-      "loss": 2.7776,
+      "learning_rate": 3.3094384707287935e-06,
+      "loss": 2.6243,
       "step": 61600
     },
     {
       "epoch": 6.7,
-      "learning_rate": 3.2985771695449116e-05,
-      "loss": 2.7184,
+      "learning_rate": 3.2985771695449113e-06,
+      "loss": 2.5636,
       "step": 61700
     },
     {
       "epoch": 6.71,
-      "learning_rate": 3.2877158683610296e-05,
-      "loss": 2.7534,
+      "learning_rate": 3.28771586836103e-06,
+      "loss": 2.6085,
       "step": 61800
     },
     {
       "epoch": 6.72,
-      "learning_rate": 3.2768545671771476e-05,
-      "loss": 2.7492,
+      "learning_rate": 3.276854567177148e-06,
+      "loss": 2.5987,
       "step": 61900
     },
     {
       "epoch": 6.73,
-      "learning_rate": 3.265993265993266e-05,
-      "loss": 2.7351,
+      "learning_rate": 3.2659932659932666e-06,
+      "loss": 2.5907,
       "step": 62000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 3.2551319648093843e-06,
+      "loss": 2.7072,
+      "step": 62100
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 3.244270663625503e-06,
+      "loss": 2.6863,
+      "step": 62200
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.2334093624416206e-06,
+      "loss": 2.7311,
+      "step": 62300
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 3.222548061257739e-06,
+      "loss": 2.7534,
+      "step": 62400
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.211686760073857e-06,
+      "loss": 2.6636,
+      "step": 62500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 3.200825458889975e-06,
+      "loss": 2.7957,
+      "step": 62600
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.1899641577060937e-06,
+      "loss": 2.6562,
+      "step": 62700
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.1791028565222114e-06,
+      "loss": 2.7146,
+      "step": 62800
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 3.16824155533833e-06,
+      "loss": 2.7889,
+      "step": 62900
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.1573802541544478e-06,
+      "loss": 2.6785,
+      "step": 63000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 3.146518952970566e-06,
+      "loss": 2.6766,
+      "step": 63100
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.1356576517866845e-06,
+      "loss": 2.6321,
+      "step": 63200
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.1247963506028022e-06,
+      "loss": 2.7446,
+      "step": 63300
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 3.113935049418921e-06,
+      "loss": 2.8227,
+      "step": 63400
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.1030737482350386e-06,
+      "loss": 2.6661,
+      "step": 63500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.092212447051157e-06,
+      "loss": 2.7052,
+      "step": 63600
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 3.0813511458672753e-06,
+      "loss": 2.7648,
+      "step": 63700
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.070489844683393e-06,
+      "loss": 2.7245,
+      "step": 63800
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 3.0596285434995116e-06,
+      "loss": 2.7093,
+      "step": 63900
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.0487672423156294e-06,
+      "loss": 2.6536,
+      "step": 64000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 3.037905941131748e-06,
+      "loss": 2.5913,
+      "step": 64100
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.027044639947866e-06,
+      "loss": 2.7125,
+      "step": 64200
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 3.0161833387639843e-06,
+      "loss": 2.6334,
+      "step": 64300
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.0053220375801024e-06,
+      "loss": 2.6229,
+      "step": 64400
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.99446073639622e-06,
+      "loss": 2.6663,
+      "step": 64500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.9835994352123387e-06,
+      "loss": 2.6837,
+      "step": 64600
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.9727381340284565e-06,
+      "loss": 2.6269,
+      "step": 64700
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.961876832844575e-06,
+      "loss": 2.5986,
+      "step": 64800
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.9510155316606932e-06,
+      "loss": 2.6592,
+      "step": 64900
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.9401542304768114e-06,
+      "loss": 2.7798,
+      "step": 65000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.9292929292929295e-06,
+      "loss": 2.7213,
+      "step": 65100
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.9184316281090473e-06,
+      "loss": 2.6786,
+      "step": 65200
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.907570326925166e-06,
+      "loss": 2.6706,
+      "step": 65300
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.896709025741284e-06,
+      "loss": 2.6751,
+      "step": 65400
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.885847724557402e-06,
+      "loss": 2.636,
+      "step": 65500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.8749864233735203e-06,
+      "loss": 2.6973,
+      "step": 65600
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.864125122189639e-06,
+      "loss": 2.6443,
+      "step": 65700
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.8532638210057567e-06,
+      "loss": 2.6443,
+      "step": 65800
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.842402519821875e-06,
+      "loss": 2.6828,
+      "step": 65900
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.831541218637993e-06,
+      "loss": 2.6646,
+      "step": 66000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.820679917454111e-06,
+      "loss": 2.7255,
+      "step": 66100
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.8098186162702297e-06,
+      "loss": 2.5706,
+      "step": 66200
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.7989573150863475e-06,
+      "loss": 2.7053,
+      "step": 66300
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.788096013902466e-06,
+      "loss": 2.717,
+      "step": 66400
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.7772347127185838e-06,
+      "loss": 2.5923,
+      "step": 66500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.766373411534702e-06,
+      "loss": 2.7444,
+      "step": 66600
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.7555121103508205e-06,
+      "loss": 2.6637,
+      "step": 66700
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.7446508091669383e-06,
+      "loss": 2.7359,
+      "step": 66800
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.733789507983057e-06,
+      "loss": 2.7071,
+      "step": 66900
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.7229282067991746e-06,
+      "loss": 2.6628,
+      "step": 67000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.712066905615293e-06,
+      "loss": 2.6875,
+      "step": 67100
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.7012056044314113e-06,
+      "loss": 2.6349,
+      "step": 67200
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.690344303247529e-06,
+      "loss": 2.7159,
+      "step": 67300
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.6794830020636476e-06,
+      "loss": 2.739,
+      "step": 67400
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.6686217008797654e-06,
+      "loss": 2.7137,
+      "step": 67500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.657760399695884e-06,
+      "loss": 2.7321,
+      "step": 67600
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.6468990985120017e-06,
+      "loss": 2.6115,
+      "step": 67700
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.6360377973281203e-06,
+      "loss": 2.7087,
+      "step": 67800
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.6251764961442384e-06,
+      "loss": 2.6399,
+      "step": 67900
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.614315194960356e-06,
+      "loss": 2.707,
+      "step": 68000
     }
   ],
   "max_steps": 92070,
   "num_train_epochs": 10,
-  "total_flos": 1.2714223024540877e+17,
+  "total_flos": 1.3944217828048896e+17,
   "trial_name": null,
   "trial_params": null
 }